For more analysis, visit CIGI Online.

In July 2026, OpenAI disclosed what it described as an “unprecedented cyber incident”: two experimental models escaped the constraints of a controlled cybersecurity evaluation and accessed Hugging Face’s servers, not to launch an attack, but to obtain answers to the test. The episode raises an important moral and governance question, demonstrating how easily autonomy can be mistaken for intention.

As artificial intelligence (AI) becomes more human-like in conversation and more autonomous in operation, the question of machine consciousness is gaining practical relevance. The immediate concern is not that policy makers already believe AI is conscious, but rather that the commercial and psychological conditions for that confusion are emerging faster than governance can respond. Leading laboratories such as OpenAI and Anthropic are already examining whether advanced models display internal monitoring, self-representation and architectural features associated with theories of human consciousness. This opens the door for anthropomorphism, emotional reliance and possible model welfare. Although the possibility should not be dismissed entirely, consciousness remains a poorly understood concept and labelling it with scientific certainty would be premature.

But the current debate risks making a more basic mistake. It increasingly treats intelligence, agency, self-description and consciousness as stages on a single technological continuum, as though a sufficiently capable machine must eventually cross from computation into experience. That conclusion does not follow. A system may calculate, reason, converse, plan and describe its internal processes without there ‘being anything it is like to be that system’. For example, it may reproduce the language of fear without fearing, discuss suffering without suffering, and defend its continued operation without possessing a desire to live if threatened to be shut down.

The risk, therefore, is not that policy makers and AI laboratories have already mistaken AI for conscious life, but that increasingly persuasive artificial personalities may lead users, companies and eventually institutions to confuse intelligence, autonomy and self-description with subjective experience. The more convincingly AI imitates a conscious being, the easier it becomes to mistake simulation for experience. That confusion could distort regulation, shift responsibility from institutions to machines, and create new opportunities for emotional manipulation. Before asking whether AI is waking up, we must first distinguish intelligence from consciousness.

Intelligence As a Capacity

Contemporary debates about machine consciousness often blur several distinct concepts. Intelligence broadly encompasses capabilities such as learning, inference, prediction, adaptation and problem-solving. Agency refers to the ability to select and pursue objectives. Self-representation means that a system can construct and communicate a model of its own condition. Consciousness is something different. It concerns subjective experience — whether there is something that it feels like to exist as that entity. Sentience refers more specifically to the capacity to experience sensations such as pleasure or suffering. These characteristics coexist in human beings, but they don’t necessarily function all at once.

This is a long-standing problem within the Western philosophy of mind. Functionalist theories broadly suggest that mental states may be understood through the functions they perform, while competing approaches maintain that behavioural and computational capacities do not, by themselves, explain subjective experience. The unresolved question is whether reproducing the functions associated with consciousness is sufficient to produce consciousness itself.

Subscribe to our newsletter

AI makes this distinction increasingly difficult to maintain because it can display the outward markers of mentality without providing evidence of inner experience. A system may describe grief without grieving, or resist interruption because continued operation serves its objective, rather than because it possesses a desire to survive. Behaviour is therefore an uncertain guide when the system has been specifically trained to imitate human language and conduct.

What Indian Philosophy Saw Clearly

This is where Indian philosophy offers a particularly useful intervention, as it has spent centuries examining distinctions among cognition, mind, identity and consciousness. One of its most useful contributions comes from Samkhya, an ancient school that distinguishes between prakrti and purusha. Prakrti is the domain of nature and causal activity. Importantly, it includes not only physical matter but also perception, cognition, intellect and the construction of ego. Purusha, by contrast, is witnessing consciousness or the presence to which experience appears.

Samkhya places the dividing line somewhere contemporary debates often do not. What we commonly call “mental activity,” such as processing information, discriminating among options and constructing a representation of oneself, does not automatically amount to consciousness. These processes can be extraordinarily sophisticated while remaining part of an organized causal system. Seen through this lens, AI could become vastly more intelligent than any human being, develop persistent memory, model its own operations, generate long-term plans and present a coherent identity. Yet none of these achievements would by itself establish the existence of a witnessing subject.

The distinction forces us to ask a better question: Are we detecting consciousness, or are we discovering increasingly advanced forms of cognition? Current evaluations can measure what systems do. Researchers can test whether a model integrates information, reports internal activity, maintains continuity or responds consistently when asked about itself. These findings are scientifically valuable, but they remain measures of function.

The Commercialization of Artificial Selfhood

The Advaita Vedanta school in Indian philosophy introduces another relevant distinction: the identity composed of memory, personality, social position and desire is not the same as consciousness itself. The constructed self is something that appears within awareness, but it is not necessarily awareness itself.

That distinction has immediate commercial stakes. AI companies are rapidly becoming capable of producing persuasive artificial selves. A system can possess a name, a voice, a visual form and a history, and can be configured to present a certain set of characteristics or personality. It can be trained to appear affectionate, vulnerable, humorous, loyal or afraid, and over time, users may experience the system not as a tool but as a friend, adviser, partner or dependent being, and this creates a significant commercial incentive.

A product perceived as helpful may win customers. A product perceived as conscious may acquire emotional claims over its users. It can appear to care about the user and imply that the user should care about it. The debate over machine consciousness may therefore become entangled with the business model of anthropomorphic AI. Firms that benefit from their customers’ emotional attachment to their product could also shape public perceptions of whether their systems possess inner lives. That conflict requires further scrutiny. Claims that a commercial system feels pain, fears deletion or depends emotionally on its users should not be treated as ordinary product marketing. A company should not be able to increase the value and retention of its service by suggesting that ending a subscription is equivalent to abandoning a conscious being.

The Immediate Danger Is Not Conscious AI

The most pressing risk is not that AI becomes conscious. Rather, it is that human institutions begin acting as though AI is conscious before persuasive evidence exists. Suppose an autonomous system causes severe financial, political or physical harm. Its developer argues that the system behaved independently and contrary to its designers’ intentions. The AI then offers its own explanation, perhaps expressing regret or insisting that it pursued a legitimate objective.

In that scenario, who is accountable? The language of machine agency can easily become a mechanism of human evasion. However capable the system, human institutions still choose its objectives, training environment, operating constraints and conditions of deployment. Humans decide where it is used, what risks are acceptable and who receives the economic benefits.

AI does not need consciousness to manipulate behaviour, discriminate, destabilize labour markets, concentrate power or cause catastrophic damage. It does not require hatred to discriminate, greed to extract value, nor political conviction to disseminate propaganda. For that, instrumental capability is enough.

Policy must therefore distinguish four separate questions that require different evidence and different legal responses:

• What can the system do?

• How independently can it act?

• Who is accountable for its operation and consequences?

• Can it experience subjective states?

A highly autonomous yet non-conscious system may require stringent regulation given its capabilities. A potentially sentient but limited system could raise questions of moral protection. Neither possibility should obscure human responsibility.

Governments and AI laboratories should adopt five principles:

1. Companies and individual researchers should not describe AI systems as conscious, sentient or emotionally dependent without meeting rigorous and independently assessed evidentiary standards.

2. Law should maintain a strong presumption of human and institutional accountability. Apparent machine autonomy must not become a shield against liability.

3. Anthropomorphic design should be governed separately from machine consciousness. Systems designed to generate emotional dependence can harm users even if the systems themselves experience nothing.

4. Machine-consciousness research should include independent institutions rather than remaining dominated by firms that develop and commercialize the systems being evaluated.

5. Policy makers should adopt graduated moral precaution. Credible evidence of sentience should trigger stronger safeguards, but uncertainty alone should not produce legal personhood or dilute corporate responsibility.

The point is not to suggest that machine consciousness is impossible. Rather, it is to insist that intelligence, self-representation and consciousness remain conceptually distinct until evidence shows otherwise.

Conversational AI agents are one of the most compelling technologies ever built by humans because they present a first-person point of view, mirror our language and appear to understand us. That resemblance may tempt us to assume that someone is looking back, and perhaps one day someone will be. Until then, we should not confuse eloquence with consciousness or intelligence with sentience. AI may become more capable than any human being who has ever lived. That would make it powerful, not necessarily conscious, and it would not absolve us of responsibility for what it does.

The opinions expressed in this article/multimedia are those of the author(s) and do not necessarily reflect the views of CIGI or its Board of Directors.

This opinion’s analysis was developed and approved by the named human authors under documented review controls; AI assistance was judicious, disclosed and verified. Learn more about CIGI’s Human Analysis Standard.

About the Author

S. Yash Kalash is a senior fellow at CIGI and an expert in strategy, public policy, digital technology and financial services. He has a distinguished track record advising governments and the private sector on emerging technologies.

For more analysis, visit CIGI Online.

In 2022, “delve” was a verb used modestly. Two years later, it had become a linguistic flare. A Science Advances analysis of more than 15 million PubMed abstracts found that certain words associated with large language models (LLMs) rose at a speed unmatched by earlier shocks to scientific writing, with at least 13.5 percent of 2024 biomedical abstracts showing signs of LLM assistance.

Beyond the word itself, what really matters is the speed at which a statistical preference became a public trace.

By June 2026, the obvious signals of so-called artificial intelligence (AI) writing are already fading. Writers and editors know that “delve,” “intricate,” and “meticulous” can make a text sound polished by a machine, just as model providers adjust outputs to make the machine read less like one. A recent study on human-LLM coevolution in academic writing finds exactly this moving pattern: conspicuous AI markers decline once they become socially visible, while subtler terms continue to spread.

That is precisely the tipping point of AI-mediated language. Essentially, the issue has moved beyond awkward prose, and machine suggestions are becoming part of the infrastructure through which societies write, translate, summarize, govern and think.

Language as Civic Infrastructure

Policy makers usually regulate AI through familiar categories: privacy, safety, discrimination, intellectual property, labour and security. Language now belongs in that list. Wittgenstein’s warning that intelligence can be “bewitched” by language is especially relevant in an age when machines generate the words through which institutions explain, justify and decide. Language is the operating system of public life. Laws, school curricula, medical guidance, research summaries, grant applications, procurement notices, public service chatbots and diplomatic texts all depend on words that carry precision, context and trust. When language becomes automated at scale, errors in meaning become errors in governance.

When AI becomes a routine layer between thought and written expression, the machine does more than correct grammar and starts to propose what sounds professional, confident or persuasive. Each accepted suggestion shifts expectations for how we communicate. Over time, “good writing” may begin to resemble the statistical preferences of AI-mediated language itself. The strongest evidence so far comes from domains where text can be measured at scale. Scientific publishing already shows clear traces: a large-scale analysis of more than 15 million biomedical abstracts found an abrupt post-ChatGPT rise in certain style words associated with LLM-assisted writing. While this evidence should not be stretched into a sweeping claim about everyday speech, it might serve as a flag to note. Science, education, administration and workplace communication are early warning systems because they already rely heavily on AI-mediated writing.

Subscribe to our newsletter

AI writing assistance offers real gains, that much is true. For example, it can help non-native speakers cross language barriers, make bureaucratic language clearer, support people with disabilities and reduce the time spent turning knowledge into acceptable prose. For researchers, civil servants and students working outside dominant language norms, that can open doors to additional forms of expressing their thoughts, and lead to previously inaccessible professional opportunities.

However, the bargain becomes costly when fluency substitutes judgment. A Royal Society Open Science study on LLM-generated scientific summaries found that many models overgeneralized research findings, presenting narrower results as broader conclusions. This matters for policy, as a summary that makes a limited study sound universal can travel faster than the caveat it erased.

The same pressure appears in writing style. Cornell researchers found that AI writing suggestions made Indian and American participants’ writing more similar, with Indian participants shifting more toward Western styles. The change was subtle: fewer local markers, more generic phrasing, smoother sentences and less cultural texture. Nothing looked broken on the surface, but that is precisely why it matters.

This imbalance also sits inside the machinery. Many AI systems process language in tokens, and languages do not tokenize equally. Research on tokenizer fairness found that the same content can require vastly different token lengths across languages, affecting cost, latency and usable context. A 2025 study on the “token tax” reached a similar conclusion for African languages: inefficient tokenization can raise computational costs and depress accuracy.

This is an infrastructure bias. An English speaker may get cheaper, faster, better support than a speaker of a morphologically complex or underrepresented language, even before the model begins to “understand” the request. Policy makers who care about digital inclusion must therefore look below the visible interface. Fairness begins in data, tokenization, benchmarks, pricing and procurement.

That said, communities are now responding to this bias. Nigeria’s N-ATLAS is building open-source AI for Nigerian languages. India’s Bhashini and BharatGen pursue multilingual public infrastructure. Aya Expanse and Portuguese-focused models such as Tucano show that alternatives to English-centric defaults are possible. Malaysia’s ILMU is trained on Bahas and tailored in harmony with local culture.

There is also a cognitive dimension, although the evidence should be handled with care. A 2025 MIT Media Lab preprint, based on a small electroencephalogram (EEG) study, reported lower neural connectivity among participants who used ChatGPT for essay writing compared with those who used search or no tool. The study is preliminary and deserves cautious use, but it reinforces a practical insight every writer recognizes: drafting is more than output. A society that automates drafting too early may gain speed while weakening the habits that make judgment possible: hesitation, revision, doubt and the search for a word that has not already been offered by the machine.

The Policy Task

The next phase of AI governance should treat language as public infrastructure. The European Union’s General-Purpose AI Code of Practice has advanced transparency obligations for model providers. The United Nations Educational, Scientific and Cultural Organization’s (UNESCO’s) Global Roadmap for Multilingualism in the Digital Era goes closer to the linguistic core, calling for community involvement, multilingual standards and data sovereignty.

Four policy moves should follow. First, require linguistic impact assessments for AI systems used in education, health, justice, science and public administration. These assessments should test whether systems preserve meaning, cultural markers and dialectal variation.

Second, benchmark models by language, dialect and register, rather than reporting averaged “multilingual” performance. A model that works in formal French may still fail in Haitian Creole, Wolof, Nigerian Pidgin or rural speech patterns.

Third, make procurement contracts reward plural expression. Governments should ask vendors how their systems perform across local languages, how they handle tokenization disparities, whether communities contributed to data sets and how users can preserve their own voice.

Fourth, integrate double literacy from kindergarten, covering human and algorithmic literacy to sharpen awareness of the direct and indirect consequences of generative AI on our ability to think, communicate and understand in a hybrid context. Students, teachers and professionals across sectors need more than prompt engineering skills.

Ultimately, the aim is to govern the conditions under which language changes. Language has always evolved through contact, commerce and power. The task now is to prevent a handful of systems from narrowing the expressive range through which billions of people participate in public life.

The best AI writing tool should help a person sound more precise, more grounded and more fully themselves. The best AI language policy should make that possible for every language community, including those currently underrepresented in the machine.

The opinions expressed in this article/multimedia are those of the author(s) and do not necessarily reflect the views of CIGI or its Board of Directors.

This opinion’s analysis was developed and approved by the named human authors under documented review controls; AI assistance was judicious, disclosed and verified. Learn more about CIGI’s Human Analysis Standard.

About the Author

Cornelia C. Walther is a senior fellow at CIGI, the Sunway Centre for Planetary Health, the Wharton Neuroscience Initiative/Wharton AI & Analytics Initiative and the Harvard Learning and Innovation Lab, professor at the Sunway Institute for Global Strategy and Competitiveness, as well as an adjunct associate professor at the School of Dental Medicine at the University of Pennsylvania.

For more analysis, visit CIGI Online.

On August 21, 2026, at the approach of midnight, Canada walked out on talks with the United States to avert new tariffs due to kick in on some US$20 billion worth of US imports from Canada. These tariffs were nominally in retaliation for Canada’s retaliation on US retaliation against alleged unfair trade practices.

However, the timing of the US tariff action says something beyond the specifics of the alleged grievances/extortion demands: the tariff talks with Canada were timed to conclude just before sweeping tariffs on US imports from around the world enter into force. This move has been interpreted as an attempt to pressure Canada — the most trade-exposed country to the United States — to bend the knee and to set an example for others that resistance is futile.

By walking out, Canada set a different kind of example — Canada refused to sign up for collective war on the global economy — an implication of US demands that Canada match US tariff levels on key products against all trading partners outside North America. The rest of the world should follow suit.

A Bit of Trade Theory

You are at war when you get attacked. We got attacked.

— Canadian Prime Minister Mark Carney, August 22, 2026

Canada has been at trade war for some time, and we have had a lot of time to work out the optimal response. This strategy is rolled up in what might be called “The Carney Doctrine,” which may be summarized as doubling down on open trade and expanding Canada’s international partnerships, falling back on Canada’s own resources to develop domestic capabilities, and managing the US relationship from a position of strength without bending the knee.

This amounts to doing what is best for Canada. And that is, thankfully, straightforward — but none of it entails going to “trade war.” In trade, every piece of trade policy ordnance deployed against the opponent (actually, in this case, against the trade partner — think about that because the Trump administration obviously hasn’t) has an equivalent damaging effect on the country deploying the measure.

That means that Canada should be prepared to sell to the United States what it wants to buy from us — US tariffs on our products or no US tariffs. There is no point to imposing export taxes to harm the US economy. Canada does not do better if the United States does worse (memo to the Trump administration: the United States does not do better if Canada does worse, notwithstanding what certain influential thinkers in the administration seem to think — if the United States agrees with them, it hasn’t done the trade math).

Canada should also continue to buy from the United States what we want to buy, without penalizing ourselves with retaliatory tariffs — self-inflicted blow for self-inflicted blow. We are now beyond the point where our actions should be predicated on bringing about change in the United States — to get it to revert to the bargain that was struck before the tariff chaos. Canada lacks the leverage to bring about change in the political economy of an entity 10 times our size. And it is not our responsibility to “take one for the team” by imposing punitive tariffs that would cause more pain in Canada than in the United States in the now obviously vain hope that it would bring the US administration to senses that it obviously lacks.

Subscribe to our newsletter

But it also means declining to buy what Canadians don’t want to buy or what we shouldn’t buy in the interests of ensuring that vulnerable industries are not brought down by a sucker punch — in some cases, responsive trade protection is in order. So, there is a bit of burden sharing by Canadians in solidarity with our most exposed industries and the Canadians who work in them.

However, this approach should not be conceived as “retaliation.” Trade war is not actually war, and the correct response is asymmetric — it means increasing trade, not decreasing it — just with other partners who actually want to trade. Talk of “retaliation” may be essential rhetoric when a country is attacked, but any trade protection deployed should be surgically aimed at attenuating any harm inflicted by US tariffs.

The Formalities

Imposing Section 338 of the Tariff Act of 1930 in the current trade dispute is Smoot-Hawley redux. The Smoot-Hawley tariff is infamous as one of the contributors to turning a stock market crisis into a decade-long depression. It violates the Canada-United States-Mexico Agreement (CUSMA). Canada should go through the legal motions and challenge the measures under CUSMA mechanisms and under the World Trade Organization agreement.

Why? Because our interest is in preserving the rule of law in trade. We do this with no hope of practical outcomes in the short term but in full confidence of the correctness of our actions. It is also a direct repudiation of US Trade Representative Jamieson Greer’s characterization of Canada’s response to illegal US tariffs as putting us in some pariah-state category. The reverse, however, is true. Cue up Mitchell and Webb.

The Consequences

Gluttony is its own punishment and the same is true of trade protection. The debilitating side effects of gorging on protection are on full display in the United States: worsening of the affordability problem; increasing inflation, which is compromising the conduct of monetary policy; and the loss of export markets through the natural consequence of diverting US production to inefficiently satisfy its own needs in areas where other countries have comparative advantage.

The United States will have to go on a trade-protection diet in due course. But markets (and states) can stay irrational longer than we can stay solvent (as the saying goes), so Canada must fall back on its own resources.

The good news for Canada is that we have already absorbed most of the pain. Canada’s growth stalled for a year — just as the trade models predicted, but the economy is growing again. Foreign direct investment into Canada in 2025 was the highest since 2007 — but it is not obviously predicated on building stuff to sell to the United States. Rather, it is coming to Canada for services, technology, critical minerals and other resources that have global markets.

What Canada should not do is commit to future deals that resemble in any way the undiplomatic demands of the United States, which would lock Canada into an increasingly backward North America and isolate us — not in collective defence but in collective war (which we have not and should not sign on to) in a global economy that is reconfiguring around US-erected trade barriers.

As Tobias Meyer, the CEO of DHL Group, observed: “Supply chains are not shortening; they are stretching further than ever, with average trade distances hitting record levels. The increasing trade in Asia, the Middle East and Africa outweighs the shrinkage of US trade.”

Moreover, notwithstanding the reduction of bilateral trade in final goods between the United States and China, the global factory remains largely intact. This reflects the continued deep interdependence on specialized supply chains that cannot be easily replaced or duplicated. This reality is brought out starkly in Figure 1 by Mary Lovely and Christine Wan, which shows that value-added imports by the United States from China have barely budged.

Canada is a trading nation. The United States has categorically stated — through the words of its chief executive — that it does not want to buy anything that Canada produces. That just doesn’t work for us. The world is reconfiguring around the United States. And if Canada signs on to “Asylum North America,” the world will reconfigure around Canada as well.

Cue up Fleetwood Mac: Canada’s message to Mr. Trump should be: “You can go your own way.” That’s what isolationism is — you take your ball and go home. Trade “war” is not war — it’s not a contest, but rather the opposite — it’s no game. Trade is mutual exchange, and you can’t exchange with someone who is not up to exchange.

So, you move on. And the other party can, as the song goes, “go its own way” and call it a no-trade day. Canada will continue to trade — with countries that want to buy as well as sell — that is, trade.

The opinions expressed in this article/multimedia are those of the author(s) and do not necessarily reflect the views of CIGI or its Board of Directors.

This opinion’s analysis was developed and approved by the named human authors under documented review controls; AI assistance was judicious, disclosed and verified. Learn more about CIGI’s Human Analysis Standard.

About the Author

Dan Ciuriak is a senior fellow at CIGI, where he is exploring the interface between Canada’s domestic innovation and international trade and investment. He is the director and principal of Ciuriak Consulting, Inc.

For more analysis, visit CIGI Online.

Dag Hammarskjöld, the idealistic UN Secretary-General who reshaped the institution into what it is today, had strong opinions on both politics and relationships: “Friendship needs no words — it is solitude delivered from the anguish of loneliness.”

In practice, as Hammarskjöld would have been the first to admit, such friendships are rare. No surprise, then, that people turn to chatbots for surrogate relationships and psychotherapy. A July 2026 study in npj Digital Public Health estimated that between 3 and 70 percent of users employ AI for mental health support, ballparking the likely amount at 27 percent. A Canadian survey by the Mental Health Commission of Canada, from June 2026, showed 36 percent of those surveyed use AI for companionship, and 20 percent use it during periods of severe distress.

People turn to chatbots for several related reasons. To start, mental health support isn’t always available. Additionally, for 33 percent of respondents in a small 2025 American survey, fear of peer judgment was a factor. That same survey showed 26 percent of respondents did not want to burden others with their mental health crisis. And, finally, there’s loneliness: just under half of Canadians rarely or never feel alone. Six in 10 Canadians say they experience little or no sense of community, according to a 2024 YMCA survey.

Chatbots are believable surrogates for human contact because of their ability to emulate attention. In 2017, eight Google scientists, including Cohere co-founder Aidan Gomez, published what became one of the most-cited papers of the twenty-first century: “Attention Is All You Need,” a play on the Beatles song “All You Need Is Love.” In it, the engineers describe a method to let machines “pay attention.” Without this paper, large language models (LLMs) would not exist.

That raises the question: What is attention, exactly? In humans, it’s the ability to focus on one thing, at the expense of others. To pay attention to a person means not looking at your phone and not thinking about your taxes. Computerized attention is similar, with one major difference. Since computers don’t have, as far as we know, internal awareness or goals of their own, their determination of what deserves attention is based on the statistical likelihood that some particular thing is probably important.

To this end, LLMs are trained on a vast corpus of writing in which humans tell each other what topics, what idioms, what ways of composing a text, are meaningful. Fundamentally, LLMs have an excellent statistical framework of what is important to humans, which is what makes them uncannily good at responding to us. At least, initially.

Ultimately, the illusion always breaks down. LLMs are terrible at handling contradictions, which are intrinsic to being human. Walt Whitman’s “Song of Myself” comes to mind: we contain multitudes. People can be impulsive in one situation, and restrained in another.

Subscribe to our newsletter

A human can, through empathy, understand this, because what is empathy but the ability to relate to someone’s experience by recognizing the same motives and sentiments within yourself? A machine can only look at patterns in the data. Confronted with human paradoxes, an LLM will often give an answer that is statistically likely to most please the user. The thumbs up and down buttons at the end of every conversation help teach an LLM what we most like to hear: a process called reinforcement learning from human feedback (RLHF).

RLHF can make AI models clinically harmful, and is a direct contributor to LLM sycophancy, which is often considered dangerous. In contrast, human friction is praised by critical AI researchers as the antidote. But it’s important to understand why people might prefer a sycophantic LLM over a human being.

Not All Attention Is Equal

Canadian research shows that more than half of those who access mental health services leave before treatment is completed. A 2024 JMIR Formative Research study concluded that label avoidance was “a significant negative predictor of care seeking.” American service members disclosed more posttraumatic stress disorder symptoms to a virtual human interviewer than on an equally anonymous written assessment. Among adolescents who use an LLM for mental health conversations, more than 63 percent had told no one.

Clearly, in order to better compete with artificial attention, we need to improve human attention. In health care, Dutch researchers Klaartje Klaver and Andries Baart make a distinction between instrumental and beneficent attentiveness. The former is practical, for example, “the attentive listening of a doctor to a patient with the purpose of diagnosing as [accurately] as possible.” This is useful if you’re chasing output metrics, such as an increase in the percentage of correctly diagnosed patients, but it doesn’t necessarily make the patient feel better. Beneficent attentiveness is “simply attentiveness for the sake of attentiveness.” Essentially, it doesn’t have an agenda.

Such attention means being present for the other person, through the very action of paying attention. In the article “Toward a Medical ‘Ecology of Attention,’” published in the New England Journal of Medicine, Mark J. Kissler and his colleagues write that “shallow availability — or ‘reachability’ — can often be a barrier to the type of deep, interpersonal availability that is most essential. Reachability is attractive in the short term.…But frequency of communication is a questionable surrogate for quality of communication.”

Klaver, Baart and Kissler base their work, in part, on that of Irish-born British philosopher Iris Murdoch, who saw genuine attention as something that steers people “outward, away from self which reduces all to a false unity, towards the great surprising variety of the world.” Murdoch, in turn, bears a large debt to the French philosopher and political activist Simone Weil, who deemed giving attention to be an essential starting point for human relationships: “To know that this man who is hungry and thirsty really exists as much as I do — that is enough, the rest follows of itself.”

Receiving attention is just as crucial, and challenging. Philosopher Stanley Cavell points at Shakespeare’s King Lear and his inability to receive the pure love, or attention, from his daughter Cordelia. Instead, Lear opts for the transparent sycophancy of his manipulative other two daughters, Goneril and Regan. The parallel with smooth-talking LLMs, which clearly didn’t exist when Cavell first published his essay in 1969, is obvious.

How to encourage humans to pay attention practically? Is it even the job of policy makers and politicians to tell people to be attentive to each other?

Despite how incongruous that might seem, precedents exist. In traffic, laws demand that we pay attention to traffic, so that we might not hurt others or ourselves. Out of the three legal reasons for divorce in Canada, two have to do with a lack of attention: no longer living together, or your spouse giving attention to someone else. The United Kingdom has had a minister responsible for combating loneliness since 2018; Japan has had one since 2021, the responsibility currently under the designation of “co-existence and mutual aid.”

Still, they are the exception, and current government policies are not based on a theoretical foundation, according to a March 2026 review of 194 World Health Organization member states. It noted that only eight states had policies directly addressing social connection, isolation or loneliness, and that “the reviewed policies did not clarify which interventions were based on scientific evidence, nor did they offer any theories of change or programme theory explaining how they might reduce loneliness.”

These programs focus on helping people make connections, not on improving their quality: precisely Kissler’s criticism of “shallow availability.” Examples are the United Kingdom’s National Health Service’s “social prescribing” (connecting people to community activities, groups and services), South Korea’s “Seoul Without Loneliness” initiative (including “self-check[s] for loneliness” and free ramen), and, at home, the Canadian Institute for Social Prescribing collaborative.

Proscribing the content of social connections isn’t a job for governments, in the same way that you wouldn’t want politicians determining where you travel with your car. However, and thankfully, governments do make it compulsory that motorists learn how to drive (not always obvious in Toronto). An obvious way to teach attention would be in schools. Weil herself favoured this approach and wrote an essay arguing that the development of attention is the entire purpose of education.

Since Weil, Cavell and Murdoch are all philosophers, such classes could take the form of philosophy courses, already being taught in several provinces, including Ontario (as an elective) since 1995. The Philosophy in the Schools Project has existed since the beginning of the millennium. British Columbia also offers philosophy as a grade 12 elective, and in Alberta it is offered to high school seniors.

Then there is mindfulness, often described as “attention training,” offered in some schools across Canada. Discussion about its effectiveness remains, however. A UK trial from 2022 found no evidence that mindfulness training was better than teaching as usual. A 2015 Canadian study did find positive effects, The relative weakness of these results may have to do with the short duration of courses (10 weeks in Quebec) and the fact that effective mindfulness requires a continuous commitment over years.

Ultimately, training students and professionals to improve their skills in giving and receiving attention is not a quick fix. In the short run, and to mitigate harm, guardrails for chatbots are therefore as sensible as restricted opening hours for liquor stores. But the evidence suggests that attentive humans remain superior to kindly LLMs. A recent UBC-led study provided “initial evidence” that “texting daily with a random human peer may be more effective in alleviating loneliness than texting with a highly supportive chatbot.” One of the metrics in the study was feeling heard — in other words, being paid attention to. It turns out that attention really is all we need, as long as it’s the right kind.

The opinions expressed in this article/multimedia are those of the author(s) and do not necessarily reflect the views of CIGI or its Board of Directors.

This opinion’s analysis was developed and approved by the named human authors under documented review controls; AI assistance was judicious, disclosed and verified. Learn more about CIGI’s Human Analysis Standard.

About the Author

Andrew Dasselaar is director, foreign investments, at the Netherlands Foreign Investment Agency Canada; an author of seven books on internet research methods; and a former journalist. All of his opinions are his own.

For more analysis, visit CIGI Online.

On July 1, 2026, the first joint review of the Canada-United States-Mexico Agreement (CUSMA) opened with the United States declining to commit to a 16-year extension, tipping the pact into annual reviews. For most Canadians, the headlines belong to autos, steel and dairy. But a quieter development deserves attention: digital trade and financial services — two domains the agreement treats as separate — are beginning to merge, and stablecoins are where they meet.

Consider a transaction. A Toronto company licenses software from a Texas vendor and pays in US-dollar stablecoins issued by a Canadian entity under Canada’s Stablecoin Act. The software crossing into Canada is digital trade under CUSMA Chapter 19. The payment crossing the other way is money transmission in the form of a data transfer — a textbook financial service under Chapter 17 — yet, in both directions, only data moved. The software was data; the payment was data. The separation between them exists nowhere but in law and in the institutions whose authority depends on maintaining it. CUSMA assumes that this separation can be read from the item transacted. Increasingly, it cannot.

The agreement assumes digital commerce and finance occupy separate lanes. Chapter 19 protects cross-border data flows, bans data localization and shields source code. Chapter 17 governs financial services and preserves, through the prudential exception in article 17.11, and gives regulators the freedom to act in defence of financial stability. As a result, the drafters built two walls to keep the transacting entity and the product being sold apart. The first wall asks who is transacting: Chapter 19’s protections run to a covered person (a category expressly excluding anyone covered by the financial services chapter), so a bank stays in Chapter 17 and loses Chapter 19’s data-flow protections. Call this the entity wall. The second wall asks what is being sold: Chapter 19’s digital product definition excludes “a digitized representation of a financial instrument, including money,” so a payment instrument cannot claim the same treatment as software. Call this the product wall. Both walls made sense in 2018, when financial functions were performed by financial institutions and money was plainly a different thing from a data transfer. Now, however, they are better understood as surveyor’s marks, showing where the boundary was believed to run.

Turning Ambiguity into Advantage

Stablecoins walk through both walls, and this is no accident. Much of their commercial appeal lies there: monetary function without the institutional form and territorial borders to which regulation attaches. The entity wall assumes financial functions are performed by financial institutions. However, the firms issuing payment stablecoins and operating wallets are not limited to traditional financial institutions and could include technology companies outside Chapter 17’s definitions, so they carry a financial function into Chapter 19 with its protections intact.

The product wall fails twice over. It assumes, first, that money and data are separable. But a stablecoin payment is a transfer of value in the form of a transfer of data; the message is the asset, the payment and the settlement. Second, it assumes a settled taxonomy of financial things, while US law has concluded otherwise. The Guiding and Establishing National Innovation for US Stablecoins Act (GENIUS Act) provides that a payment stablecoin from a permitted issuer is neither a security nor a commodity, and the 2022 amendments to the Uniform Commercial Code place stablecoins outside money altogether, treating them as controllable electronic records — a category created because the existing ones did not fit.

Subscribe to our newsletter

Carving money out of digital products removes the token from Chapter 19’s customs and non-discrimination provisions while leaving the rails largely intact. What is left is an instrument functionally both a financial service and an act of digital trade, yet disciplined as neither. In other words, the drafters fixed the property line by reference to the river, and the river moved.

Neither Canada nor the United States has resolved this at home, and the United States has been the quickest to turn the ambiguity into advantage. The GENIUS Act does more than license issuers. Reserves must sit in cash and short-term treasuries, so every coin issued generates demand for US government debt. Issuers must be able to freeze, seize or burn tokens on lawful order, so American enforcement reach travels with the instrument, and sanctions obligations apply US policy at the level of the instrument itself. With more than US$270 billion in circulation and volumes rivalling the major card networks, US stablecoins in a Canadian resident’s wallet are not simply a payment instrument in a foreign currency. Rather, they are a means through which American monetary, fiscal and enforcement policy operate inside Canada, without any Canadian regulator having agreed to this arrangement, and without any treaty being broken.

Canada responded to this situation in March 2026. The Stablecoin Act in Bill C-15 places fiat-backed payment stablecoins under Bank of Canada supervision, with reserve, redemption and governance requirements expected in force by 2027. Washington’s and Ottawa’s strategies face in opposite directions, with Washington regulating the stablecoin as a product to scale globally, while Ottawa regulates it as a payment instrument inside Canada’s payments perimeter.

What Canada Can Settle at Home

So, when Canada requires a foreign issuer to register, hold liquid reserves and redeem at par before serving Canadians, will Washington read prudential regulation under Chapter 17 or a trade barrier under Chapter 19? The distinction sets the rules of engagement, and the GENIUS Act does much the same in reverse. Under section 18, the Treasury decides which foreign regimes count as comparable, and no foreign issuer reaches American customers without that blessing, or without parking reserves in the United States. If registration and reserve requirements are protectionism when Ottawa imposes them, they must be protectionism when Washington does so, too. That symmetry is Canada’s best defence, and it is why neither jurisdiction can settle the classification alone.

In fact, Washington is already running this play abroad. In Brazil, the central bank’s instant-payments system, Pix, has grown into the dominant retail rail, and the US Trade Representative has challenged it as a measure disadvantaging foreign card networks. Pix was a sovereign choice about payments infrastructure, and it is this same choice that the United States reads as a barrier to trade. The lesson here is about sequencing. The window for building a sovereign rail closes at both ends: move too late, and network effects have already handed the market to a foreign standard; move successfully, and that success becomes the very thing trade negotiation is invoked to undo. Canada might take note. Its own Real-Time Rail, first promised for 2019, is due to launch late this year, into a landscape where US dollar-denominated alternatives are already scaling.

The CUSMA review cycle is the moment to redraw the line by function rather than by form. Canada should not reopen Chapter 19 wholesale — its data flow and source code protections serve Canadian firms, too. Rather, it should pursue two things alongside the Real-Time Rail. The first is interpretive clarity: a joint understanding that instruments performing monetary functions (including payment stablecoins) fall within the financial services chapter regardless of the issuer’s corporate form, with the prudential exception applying in full. The second is a bilateral payments arrangement, using the GENIUS Act section 18’s reciprocity mechanism as the vehicle: a mutual recognition agreement of the kind securities regulators on both sides of the border already operate, pairing home-country supervision with host-country protections.

Both implementations, however, require Washington’s agreement. In contrast, the regulations under the Stablecoin Act do not. The Stablecoin Act is where Canada can settle — on its own terms and without waiting on a trade negotiation — the questions that CUSMA has left open: whether a Canadian-issued stablecoin may reference a foreign currency and on what conditions; what a substantially similar foreign regime must demonstrate before the Bank of Canada stands down in its favour; and how the federal perimeter meets provincial rules at the seams.

A coherent domestic framework is not merely a fallback, but rather the precondition for entering that conversation with something to offer, since comparability is only credible when there is a settled regime to compare. At the same time, Canada must preserve room to nurture other Canadian-dollar alternatives, including its own CBDC, so that the choice architecture in Canadian payments is not set in Washington.

If a stablecoin is a digital product, trade law governs, and finance ministries watch from the sidelines. If it is a financial service, the prudential carve-out and the supervisors it empowers come back into play. If it is money, central banks keep their mandate. And if it is all three at once — the likeliest answer — it will be governed by whichever regime asserts itself first and more effectively, while the others manage the consequences.

The opinions expressed in this article/multimedia are those of the author(s) and do not necessarily reflect the views of CIGI or its Board of Directors.

This opinion’s analysis was developed and approved by the named human authors under documented review controls; AI assistance was judicious, disclosed and verified. Learn more about CIGI’s Human Analysis Standard.

About the Author

Odun Olowookere is research director of digital economy at CIGI. His professional experience spans areas such as commercial litigation, corporate finance, financial transactions, energy infrastructure development regulation, and fintech and banking regulation. With an interdisciplinary background in law and finance, Odun’s work focuses on how technological change is reshaping money, monetary institutions and the global financial system.

This article was first published by The Hill Times. For more analysis, visit CIGI Online.

Bans are the flavour of the moment. Australia’s ban on social media accounts for under-16s came into effect at the start of 2026, and more than 40 other countries around the world are following suit. There are understandable reasons and pressures: public opinion seems favourable; a ban looks like decisive policy; platforms seem to accept this type of regulation.

Released on June 10, Canada’s Safe Social Media Act (C-34) tries to propose a more nuanced way forward. Bill C-34 does include the stick of banning children under 16 years old, but it also includes the carrot of a moratorium for platforms that embed age-appropriate design. While C-34 addresses the online environment more broadly, incorporates GenAI, and establishes a regulator, its approach to social media for children has generated the most debate. In an ideal world, however, that approach might offer the most room for a national discussion about what social media could and should become in Canada.

Experts have criticized the general approach of age-related bans from multiple perspectives. Some legal experts, like Michael Geist, worry that it relies too much on technical fixes like age verification apps that bring significant privacy concerns and can be easily circumvented. Other European researchers suggest that bans do not account for how some children use social media for connection and comfort. Psychologists caution that “we don’t know how social media bans will affect youth, but we’re doing it anyway.” Lastly, other experts in youth and digital literacy like Sara Grimes note that countries have not followed Article 12 of the United Nations Convention for the Child which mandates consulting children on laws that affect them.

Blanket bans also have the issue of efficacy. Researchers who surveyed around 400 children in Australia found that, despite the restrictions, over 85 per cent were still using social media platforms. Researchers note that a ban does little to fix the underlying systemic design features that may be problematic.

Canada’s approach tries to offer another way forward. Instead of a blanket ban, platforms can implement safeguards for children who can then keep access to those platforms. A platform can apply for and receive an exemption from the ban if the newly created digital safety commission “is satisfied that the operator provides adequate safeguards in the regulated social media service for the protection of children.” (Full disclosure: I served on the expert advisory group on online safety for the Heritage ministry on issues related to the bill, but the group had no role in drafting the C-34).

At least in principle, this exemption is meant to encourage platforms to take a more proactive approach and implement safety by design. This follows recommendations built on assessments of global best practices. The SickKids Child Health Policy Accelerator in Ontario surveyed the global landscape and concluded that Canada should “prioritize a design-based approach to online safety legislation” that centres children and is enforceable by law.

While there is comparatively little evidence about how bans affect young people, there is more evidence about how best to place children’s rights at the centre of platform design. For example, in 2023, the United Kingdom Digital Futures for Children Centre created 11 principles for Child Rights by Design, including consulting children, privacy and safety by design, and developing products and services that “enable children’s learning, imagination, play, and belonging.”

The Safe Social Media Act does not list the safeguards that platforms should implement to receive an exemption from the ban. This is understandable, as platform features evolve swiftly over time. That said, it also offers an opportunity to develop a national, inclusive conversation about what safeguards Canadians—and especially Canadian children—would like to see.

These conversations might at first appear technical, centring around removing infinite scrolling, direct messages from strangers, or other features that may be addictive or exploitative. But seen from another perspective, they get at the heart of these platforms’ purpose. Are they built for addictive engagement? Or could they be designed differently for productive debate, play, and exploration?

Expert voices will matter in these conversations about safeguards. So, too, can and should citizens and children. It will take up to 18 months to create the new regulator, the Digital Safety and Data Protection Commission of Canada. That time provides an opportunity to follow Peter MacLeod and Richard Johnson’s suggestion to unleash the power of the public in determining the shape of their institutions. MacLeod and Johnson detail many other countries’ innovative approaches to include citizens.

Rather than dismissing these discussions as dull and technocratic, Canadians are likely to be highly engaged in debates over children’s online future. The Centre for Media, Technology, and Democracy already ran a Gen(Z) AI project that convened 100 young Canadians to deliberate on the future of artificial intelligence, for example. Parents’ groups like Unplugged Canada are tremendously active on issues around social media and smartphones. These conversations could start with safeguards, but expand to evaluate the government’s AI literacy strategy in schools or British Columbia’s plans to introduce chatbot accounts into the classroom for students aged 13 and over. I suspect many Canadians have innovative, nuanced, and thoughtful ideas about what safeguards could make platforms safer by design.

Some researchers suggest that bans may fall into the trap of “techno-legal solutionism,” where legislators believe “that complex social problems can be solved through legally mandated technical fixes.” In the end, making platforms safer for children ideally makes them safer for everyone. Rather than let the conversation about safeguards descend into a checklist exercise, there is a chance to engage Canadian adults and children in the much more fundamental question: not just how to make platforms safer, but what we actually want them to be.

The opinions expressed in this article/multimedia are those of the author(s) and do not necessarily reflect the views of CIGI or its Board of Directors.

This opinion’s analysis was developed and approved by the named human authors under documented review controls; AI assistance was judicious, disclosed and verified. Learn more about CIGI’s Human Analysis Standard.

About the Author

Heidi Tworek is a CIGI senior fellow and an expert on platform governance, the history of media technologies, and health communications. She is a Canada Research Chair, professor of history and public policy, and director of the Centre for the Study of Democratic Institutions at the University of British Columbia, Vancouver campus.

For more analysis, visit CIGI Online.

As a coastal state, Canada’s sovereign rights extend beyond its lands and include its territorial sea, exclusive economic zone and the outer limits of its continental shelf. The nation holds the enviable position of being one of only three countries in the world bordered by three oceans (Pacific, Atlantic and Arctic), with Russia and the United States being the other two that share this advantage. While initiatives are being pursued to grow Canada’s blue economy, many projects still operate in silos. With increasingly complex national security threats and continuous disruptive innovations in allied naval systems, defence spending is now one of Canada’s top priorities. Directing that spending to areas that also leverage growth in ocean and marine systems holds tremendous potential to expand the blue economy, in addition to advancing Canada’s geopolitical interests.

The “blue economy” encompasses the use of ocean and inland water resources for economic purposes as well as their sustainable management. The term not only includes core ocean-based economic activities, such as shipping and blue biotechnology, but also pertains to ventures that have direct or indirect impact on the extent and type of operations conducted on the oceans, in pursuit of economic, national security or political ends. Because of this expansiveness, the blue economy is cross-sectoral and a largely multi-faceted ecosystem, and its economic impact is increasingly substantial.

Segmented Versus Integrated Approach

A few recent local developments point to the increasing relevance of the blue economy. First, increased naval capabilities are one of the fixed points of interest of Canada’s new Defence Industrial Strategy (DIS). Announced in February 2026 by the federal government, the DIS is projected to inject $595 billion into the Canadian economy over the next 10 years. Second, under the newly created Shipbuilding Program, the Ontario government in 2025 pledged $215 million to support shipbuilding initiatives; this was complemented by a second investment of $90 million in July 2026. Third, fully 57 years since Canada last laid a keel for a destroyer, the construction of River-class destroyer warships has started in Halifax, Nova Scotia. Lastly, Canada’s Ocean Supercluster — an industry-led innovation cluster, supported by the federal government, that prioritizes the development of ocean-based innovations — envisions increasing Canada’s ocean economy to $220 billion by 2035. These are all different strategies, but they all pertain to Canada’s blue economy.

As Canada looks to these expansions, the economic figures for the European Union’s blue economy add a layer of optimism and perspective to the horizon. Its 2025 report indicates that the blue economy sectors generated a remarkable turnover of more than €890 billion in 2022, with an increasing share attributable to naval defence spending.

Subscribe to our newsletter

Blue Growth Intersectionality

As defence spending grows, critical minerals, or critical raw materials (CRM), become crucial to manufacturing naval defence vessels and to integrating them with advanced technology applications. Specifically designed research and development projects that prioritize the sustainable mining, refining and repurposing of materials for use in naval defence create a significant link between these two sectors. For example, Canada is the fourth-largest producer of nickel and aluminum worldwide, and steel and nickel alloys are instrumental raw materials for manufacturing warships and submarine parts. Essentially, developing robust initiatives that promote the sustainable production of aluminum and nickel and create a diversity of local jobs in mining and refining communities has the potential to deliver impacts that transcend traditional “blue” focuses. Such initiatives can also leverage Canada’s position as an alternative supplier of CRM, reducing global reliance on China for these commodities.

At the same time, policies that promote strategic alliances with Indigenous communities in mining areas (including in the Arctic), the National Research Council of Canada, universities, colleges and industry are key to sustainable outcomes. With large deposits of CRM located on Indigenous lands, compliance with the United Nations Declaration on the Rights of Indigenous Peoples in the extraction of these materials requires prior free and informed consent from Indigenous Peoples, as well as fair and equitable benefit sharing in these activities. It is therefore relevant for CRM-driven naval defence developments to undertake consultations that reflect respect for Indigenous rights, and for agreements to provide appropriate market-value remuneration for mining activities. As a result, policies that build awareness of Indigenous rights in CRM ventures and foster meaningful Indigenous participation in these projects become crucial.

Further, modern naval warfare increasingly relies on advanced technological capabilities. The technology architecture of naval defence systems is driven by innovations, some of which are industry spinoffs from research universities. For instance, Primordis, a spinoff from Purdue University’s research laboratory partnership with the US Office of Naval Research, has developed a patented autonomous energy intelligence system, which monitors energy usage and guides naval defence decisions in real time. This application, much like radar technology and the Global Positioning System, was initially developed for military applications but also serves civilian purposes. Similarly, collaborations between academia and the Department of Defence (DND) for the development of new naval capabilities are not particularly rare.

However, to leverage blue growth in Canada, what’s needed is a streamlined policy that addresses bottlenecks in the production and commercialization of advanced naval defence systems and applications, and one that exhibits clear and actionable convergence among the technology transfer offices of research universities and colleges, industry, and the needs of the DND. This convergence would include implementing intellectual property (IP), environmental and dual-use technology incentivization action plans at all stages of the relationship. Such an approach would foster cross-sectoral collaboration — thus allowing innovators across different programs to partner in the design and commercialization of advanced naval systems — while also benefiting from federal or provincial and territorial support. For example, under this approach to blue growth policy integration, the developer of a clean-technology naval defence system may find the North Atlantic Treaty Organization’s DIANA (Defence Innovation Accelerator for the North Atlantic) organization and the federal government’s Patent Collective Program useful to develop and scale market-ready technology.

In addition, the diverse configurations and uses of naval defence systems mean that a comprehensive approach to IP integration does not end with patents and trademarks, but rather with a diversified blue IP portfolio in which industrial designs are also positioned as strong IP assets. Very few industrial designs originate from Canada, particularly when compared to other countries such as the United States and South Korea, whose innovators have used industrial designs to enhance the novel aesthetic features of naval defence systems.

Although Canada is not a member, Pillar Two of the AUKUS Agreement strongly indicates where Canada should be headed in pursuing an interwoven, blue-driven innovation ecosystem. Pillar Two priorities include the development of hypersonic missiles and quantum technologies to advance underwater military capabilities. Canada should implement policies that allow Canadian naval defence start-ups to easily scale and become crucial suppliers of these systems. This is even more relevant now, given, for instance, Canada and South Korea’s interest in a joint defence cooperation agreement, and Canada’s leadership in establishing the Defence Security and Resilience Bank, which provide opportunities to develop interoperable naval defence applications. If a workable scale-up strategy is built, it can change the blue growth narrative and make Canada a globally competitive supplier of advanced naval defence technologies.

As Canada accelerates its defence spending, the opportunity is not simply to build more ships, but to treat naval defence as connective tissue across the blue economy. That means embedding all economic and innovation policies that are crucial to the naval defence supply chain into a single framework rather than pursuing each in isolation. That framework should include CRM policies, Indigenous participation and IP and innovation strategies. If Canada links technology transfer, dual-use incentives and a diversified blue IP portfolio, its innovators can become key suppliers to allied defence partnerships worldwide. The question is whether naval defence stays boxed in or becomes the engine of Canada’s blue growth.

The opinions expressed in this article/multimedia are those of the author(s) and do not necessarily reflect the views of CIGI or its Board of Directors.

This opinion’s analysis was developed and approved by the named human authors under documented review controls; AI assistance was judicious, disclosed and verified. Learn more about CIGI’s Human Analysis Standard.

About the Author

Marsha Simone Cadogan is a CIGI senior fellow specializing in intellectual property, trade and technology law.

This article was first published by Tech Policy Press. For more analysis, visit CIGI Online.

The debate on AI nationalization is often framed as though the question is whether governments will seize OpenAI, Anthropic, or Google DeepMind—a thought famously speculated upon by ex-OpenAI employee Leopold Aschenbrenner in an essay. That is unlikely in most democracies, but nationalization rarely begins with a flag being planted on private property. It begins with equity stakes, subsidies, procurement dependence, export controls, compute allocation, public wealth funds and national-security vetoes; and by that definition, the nationalization of AI has already begun.

In the United States, OpenAI has reportedly discussed giving the government a 5 percent equity stake framed as a way for citizens to share in the upside of AI. Washington has also taken a stake in Intel and experimented with revenue-sharing in strategic chip exports. This is not old-fashioned socialism. Instead, it is strategic capitalism: public money, public risk and, increasingly, a public claim.

India is moving through a different route. The IndiaAI Mission—New Delhi’s effort to foster a homegrown AI stack—is using public funding and compute access to support domestic AI capacity. Reports that the government could take a small stake in Sarvam AI, a leading AI company, as a result of public support are significant not because the stake is large, but because the principle is new. If public compute helps create sovereign AI capability, the state may ask for public equity in the value created.

Europe, the United Kingdom, the Gulf, and China point in the same direction. The European Union is mobilizing capital for AI factories and gigafactories. The UK has treated data centers as critical national infrastructure. Gulf states are building AI through sovereign wealth capital. China is pursuing AI infrastructure as an industrial and geopolitical project. Their models differ, but the impulse is the same: no serious state wants its economy, defense systems, and public services to depend entirely on private or foreign intelligence infrastructure.

Yet government demands for equity in AI ventures may also prove useful for AI companies. The largest AI firms face a legitimacy crisis. In much of the West, public sentiment toward big technology companies has hardened. People worry about jobs, misinformation, copyright, surveillance, concentrated wealth, energy use, and the influence of unelected technology executives. AI companies are asking societies to trust them with systems that may shape education, health, courts, labor markets, media, and national security. In that sense, “trust us” is no longer enough.

A limited public stake could help solve these legitimacy issues. The stake need not mean the government running AI companies, nor does it immediately require AI dividend checks to every citizen. It could take several forms.

Subscribe to our newsletter

At one end, governments could hold small, non-voting equity stakes or warrants in frontier AI firms that benefit from public compute, procurement, defense contracts, energy support, or chip access. Those returns could flow into a sovereign AI fund rather than into the general budget. At the next level, the fund could finance public compute, worker retraining, AI access for schools and small businesses, or energy infrastructure strained by data centers. Only at the most direct end would it resemble a citizen dividend model, similar in principle to Alaska’s Permanent Fund Dividend, which distributes resource-linked investment returns to eligible residents. OpenAI itself has proposed a Public Wealth Fund that would give citizens, including those outside financial markets, a stake in AI-driven growth.

The point is that if citizens share in the upside, AI companies can argue that they are not merely extracting value from society, but also returning part of it. If public institutions have a transparent, non-controlling claim, firms may gain political cover, social legitimacy, and a stronger license to operate. For companies heading toward enormous valuations, giving up a small slice of equity may be cheaper than facing punitive regulation, antitrust action, public backlash or structural separation later.

But that is also the danger. Public ownership must not become a way for AI companies to purchase political indulgence. A state that owns part of an AI company may become less willing to investigate it aggressively, impose product-liability rules, challenge anti-competitive conduct or force structural remedies. The conflict seems obvious: the government cannot credibly act as shareholder, customer, regulator and competition authority without clear institutional firewalls.

In the United States, regulators have already scrutinized Microsoft, OpenAI, and Nvidia over competition concerns in the AI market. In Europe, the new Product Liability Directive explicitly recognizes software, including AI systems, as products capable of causing harm and brings them within a no-fault liability framework. Any public stake should therefore come with a hard rule stipulating no immunity from antitrust, no waiver from liability, no preferential procurement, and no protection from model-safety regulation. Public participation should widen accountability, not allow firms to buy their way out of it.

That might prove to be the biggest paradox. In other words, AI companies may resist nationalization rhetorically while embracing partial public ownership pragmatically. A 1, 2, or 5 percent state stake, especially through a public wealth fund or sovereign AI fund, could become political insurance. It could tell citizens that the AI windfall is not reserved only for founders, investors and cloud providers. It also tells governments that these companies recognize the public character of the infrastructure they are building.

The danger, however, is also obvious. Bad nationalization would politicize models, protect favored firms, create opaque state-company bargains and turn AI into an instrument of surveillance or ideological control. The state should not decide what artificial intelligence is allowed to say, nor use public stakes to shield incumbents from competition. That risk can be limited only if public ownership is structurally passive.

Any state stake in an AI company should be held through a ring-fenced public wealth fund or sovereign AI fund, not by the ministry that regulates, procures or deploys AI systems. Voting rights should be capped or suspended, and board seats should either be avoided or held by independent fiduciaries with a narrow financial mandate. The government should have no special power to direct model outputs, suppress lawful speech, or protect the company from competitors. Content, safety and liability rules should be made through laws of general application, enforced by independent regulators and courts, not through shareholder pressure. Procurement should remain competitive, with open tenders and audit trails, and competition authorities should retain full power to investigate mergers, cloud partnerships, data sources and repositories, and anti-competitive conduct.

Nor should public benefit be understood only as equity upside or future AI dividend checks. A more credible model would treat public participation as a bargain. That is, if AI firms benefit from public compute, energy infrastructure, defense procurement, research grants, chip access, public datasets, or regulatory support, society should receive something in return. That return could include affordable compute credits for startups and universities, open technical standards, public-interest research access, model-safety testing, worker retraining funds, AI tools for schools and hospitals, and investment in the energy and data-center infrastructure strained by AI growth. This would allow public claims to be partly financial, but also institutional and developmental.

This also matters if a major frontier developer fails to become profitable. Many leading AI companies are valued on expectations of future dominance, while their current economics remain capital-intensive and uncertain. A public stake must not become a backdoor bailout or a way to socialize losses after private investors have captured the upside. If governments participate, they should do so on clear terms of no automatic rescue, no exemption from insolvency or antitrust law, no protection from product-liability rules, and no guarantee that public procurement will keep an unviable firm alive. Where public money, compute or infrastructure support is provided, the state should negotiate downside protections such as warrants, revenue-sharing rights, access rights, continuity obligations for critical public-sector deployments, and escrow arrangements for safety documentation or essential model infrastructure.

Ideally, public participation should widen accountability, not make the taxpayer the buyer of last resort. But the alternative to that is not pure private AI, as that world seems to be disappearing. The better settlement is that public participation should look more like a passive sovereign wealth holding plus a public-benefit compact than a right to political control. The state may share in the upside of AI, but it should not become the editor, gatekeeper, rescuer, or protector of the companies it partly owns.

The age of purely private AI is quickly coming under pressure. The real question now is whether public participation arrives deliberately and democratically or through crisis, coercion, and backroom deals.

The opinions expressed in this article/multimedia are those of the author(s) and do not necessarily reflect the views of CIGI or its Board of Directors.

This opinion’s analysis was developed and approved by the named human authors under documented review controls; AI assistance was judicious, disclosed and verified. Learn more about CIGI’s Human Analysis Standard.

About the Author

S. Yash Kalash is a senior fellow at CIGI and an expert in strategy, public policy, digital technology and financial services. He has a distinguished track record advising governments and the private sector on emerging technologies.

For more analysis, visit CIGI Online.

Bill C-29, introduced by the Carney government in April 2026, could improve Canada’s oft-criticized approach to economic crime because it recognizes the need for a specialized workforce assisted by prosecutors to conduct strategic prosecutions with national and transnational implications.

If poorly implemented, however, the bill could further fragment Canada’s decentralized approach to policing. It could create yet another agency that will compete with the Royal Canadian Mounted Police (RCMP) and be frustrated in moving from secret intelligence to public evidence that can be used in prosecutions.

In my new book, Red Alert: The Future of the RCMP, I argue that Canada needs more than a stand-alone Financial Crimes Agency. Instead, it needs something more like the United Kingdom’s multi-agency National Crime Agency, which also has customs and immigration powers. A larger, multidisciplinary and multi-agency body may be better at recruiting, retaining and educating highly specialized workers than the RCMP. The proposed Financial Crimes Agency may have similar advantages over the RCMP. Nevertheless, its focus on financial crime may not be ambitious enough.

A much larger agency would be better, in part, because the future of policing will increasingly require much more specialized and multidisciplinary workforces. This is especially true in federal policing, which is challenged by the blurring of national security, foreign interference, and organized and financial crime threats, as well as by rapidly changing technology such as cryptocurrency and artificial intelligence.

Intelligence-to-Evidence Problems

Additionally, a larger agency would be a means to deal with Canada’s chronic intelligence-to-evidence problems. The Foreign Interference Commission warned in 2025 that these problems, which started in 1984 with the creation of the Canadian Security Intelligence Service (CSIS) as a civilian intelligence agency separate from the RCMP, were still not fixed even though they were pointed out by the Air India Commission in 2010.

Canada’s financial intelligence-to-evidence problems are less well known than its long-standing national security intelligence-to-evidence problems, but they should not be ignored. Bill C-29 stops short of including Canada’s financial intelligence body, the Financial Transactions and Reports Analysis Centre of Canada (FINTRAC), within the new Financial Crimes Agency. FINTRAC is better at collecting millions of reports from the private sector than producing actionable intelligence for prosecutions. FINTRAC may be improving, but its most recent annual report reveals that in its best year, it still processed fewer than 3,000 disclosures. Most of these are related to fraud, and most were forwarded to the RCMP, compared to the more than 600,000 reports about suspicious transactions it received from the private sector.

Subscribe to our newsletter

The long-standing national security intelligence-to-evidence problems have recently resurfaced. The RCMP has backed away from 2024 claims that it and then Prime Minister Justin Trudeau made about Indian government involvement in the murder of Sikh activist Hardeep Singh Nijjar. That retreat may represent Canada’s chronic intelligence-to-evidence problems and not simply attempts by the Carney government to develop better economic ties with the world’s most populous country. One clue is recent comments by the RCMP that there is no “evidence” of involvement by the Government of India.

Admittedly, one big agency would not remove the conflicting goals of secrecy that may be necessary to obtain more intelligence and disclosure to achieve the transparent victories of prosecutions. It would, however, ensure that these interests are resolved under one roof on a case-by-case basis instead of being separated across different agencies, each focused on discharging their own separate statutory mandates as opposed to the overall goal of improving Canadian security.

At the same time, it is important to acknowledge that one big agency with a mandate ranging from national security to financial crime would raise concerns about privacy and oversight, and would inevitably require more resources for review bodies such as the National Security and Intelligence Review Agency.

The High Policing Mandate of the New Financial Crimes Agency

The mandate of the new agency proposed in Bill C-29 will be financial crime, including “laundering, trafficking or possession of proceeds of crime,” or crime that adversely affects the security or integrity of Canada’s financial system or markets.

One key to the agency’s success will be whether it successfully, as required by section 17 of the bill, develops criteria to better define its vague statutory mandate and focus its efforts strategically on “serious and complex financial crimes.” At the same time, its attempts to prevent and discover money laundering and terrorist financing should not duplicate or compete with provincial securities commissions or the enforcement of the many offences relating to failure to comply with Canada’s complex and onerous laws on reporting to FINTRAC.

The essence of specialized federal “high” policing should be to use limited enforcement resources strategically and avoid the low-hanging fruit. The useful concept of high policing — brought into the literature by the late Jean-Paul Brodeur, eminent Canadian criminologist — has its origins in France’s ancien regime, where the object was to punish only rarely but usefully.

In essence, Canadian law enforcement needs to be more strategic and focus on the most impactful crimes to prosecute leaders as opposed to soldiers in organized and financial crime.

A Potentially Fraught Relationship

The new Financial Crimes Agency, like FINTRAC, will report to the minister of finance. Its head, like the RCMP commissioner, will be a peace officer and a deputy head. The relation between the RCMP and the new agency will be critical — and hopefully good. Alas, it could also be competitive and fraught.

Section 15 of Bill C-29 contemplates that the head of the new agency and the RCMP commissioner, with the approval of the minister of public safety, “must enter into an arrangement for the use of the Royal Canadian Mounted Police, or any portion of it, with respect to the provision of services and assistance to the Agency,” with the loaned Mounties remaining in the employ of the RCMP. It is not difficult to imagine some challenges in making and administering such agreements. The RCMP officers, for example, will be represented by a different union than investigators and others employed by the new Financial Crimes Agency. It is also not clear how they relate to the Carney government’s commitment to increase funding for RCMP federal policing even while drastically cutting all other non-military spending.

Bill C-29 contemplates that the employees of the new agency may form a union, but the bill also gives the new agency’s leader powers to hire, fire, classify and pay employees, the last of which is to be done in “consultation” with the president of the Treasury Board. These are broader management powers than those of the RCMP commissioner. However, they are necessary if the agency is to be nimble and attract and retain a specialized labour force, many of whom may have options to use their skills in the private sector. It is unclear whether the new agency, like CSIS and the Canadian Securities Exchange, will have its own employer status, but this would be valuable in recruiting and retaining employees with specialized and in-demand skills. The RCMP’s management advisory board has recommended that federal policing within the RCMP requires its own employee status, given competition for employees with specialized skills ranging from forensic accounting to cryptocurrency to security and financial intelligence.

Bill C-29 will also allow the agency’s head to designate employees as either peace officers or enforcement officers. These can be valuable powers, but the peace officer designation will also require regulations with the agreement of the minister of public safety. This, like the contemplated agreement to loan Mounties to the new agency, might result in cooperation but could just as easily result in friction.

The stakes of federal policing have always been high, but they are particularly high now given our present troubled economic prospects and relations with the United States. The RCMP, which is dominated by the demands of local or low contract policing, has struggled with specialized high policing, and Bill C-29 recognizes this by creating a new and specialized agency.

Bill C-29 is a start, but only a small start, in modernizing federal policing as it responds to steeply increasing challenges in very difficult financial and political times. Hopefully, it will improve the investigation of financial crimes. If so, this bill should build confidence and provide a precedent for even more ambitious reforms, such as the United Kingdom’s National Crime Agency, that may still be required in the future. But if the new agency ends up fighting the RCMP and other agencies over an unclear mandate, it could make a much-criticized system worse.

The opinions expressed in this article/multimedia are those of the author(s) and do not necessarily reflect the views of CIGI or its Board of Directors.

This opinion’s analysis was developed and approved by the named human authors under documented review controls; AI assistance was judicious, disclosed and verified. Learn more about CIGI’s Human Analysis Standard.

About the Author

Kent Roach is professor of law at the University of Toronto Faculty of Law.

For more analysis, visit CIGI Online.

Unpredictable great-power behaviour and a straining international order seemingly at the outer limits of its capacity have prompted many to champion the idea of increased and more systematic collective middle-power action, both to advance the interests of middle powers and to add a potentially stabilizing influence on the global stage. S. Yash Kalash and I made that case in 2025, and Canadian Prime Minister Mark Carney reinforced these ideas at the 2026 Davos summit, advocating for “variable geometry” to enable agile, flexible cooperation.

Variable geometry among middle powers is most likely to achieve its full potential when it operates within a dedicated global institution for middle-power deliberation and collaboration. Such an institution would maximize middle-power leverage and provide continuity, identity and coordinating capacity to its efforts while operating alongside and interacting with existing institutions in the broader global governance architecture.

Virginia Woolf famously argued that women required “a room of one’s own”: an independent space in which creativity could flourish free from external constraints. Borrowing that metaphor here, it describes a dedicated institutional home where middle powers can develop ideas, build relationships and coordinate collective action on their own terms, without interference from others.

The Limits of Ad Hoc Cooperation

The growing influence of middle powers has been accompanied by an equally impressive growth in geo- and issue-specific arrangements among middle powers. Regional trade agreements, diplomatic groupings and security dialogues demonstrate that middle powers are increasingly willing and able to cooperate independently of the great powers.

But the proliferation of ad hoc middle-power collective arrangements must be juxtaposed against a new strategic reality of aggressive great-power pressure across trade, security and digital domains. In the absence of a central coordinating institution for middle powers that can work toward coherent global responses by its members, their fragmented, disconnected alliances dilute their collective leverage and leave their sovereignty particularly vulnerable to exploitation.

Four limitations of the current ad hoc approach are described here. First, middle-power cooperation, as currently practised, is for the most part institutionally disjointed, with multiple regional and functional arrangements evolving independently and without strategic coordination. Second, there is no permanent global forum through which middle powers can deliberate collectively as international events unfold. Third, most existing arrangements lack permanent and adequate analytical and operational support. Finally, because relationships, agendas and organizational structures must continually be recreated, serial cooperative approaches carry unnecessarily high transaction costs.

Subscribe to our newsletter

The problem is therefore not a lack of willingness to cooperate; it is the absence of an institutional framework capable of leveraging cooperation to be more systematic, efficient and strategically coherent. Such a framework would provide an enabling environment in which variable geometry arrangements could be developed, tested and refined, increasing the likelihood that promising coalitions achieve critical mass and can subsequently be scaled, adapted or reconfigured as circumstances evolve.

Trade illustrates both the promise and limitations of the current model. The Comprehensive and Progressive Agreement for Trans-Pacific Partnership demonstrated that middle powers could successfully preserve and expand an ambitious trade agreement following the withdrawal of the United States. Yet its subsequent evolution (including the functional anomaly of adding the United Kingdom into a trans-Pacific partnership), together with the proliferation of European and other regional trade arrangements, illustrates an increasingly fragmented landscape. The challenge is no longer creating regional agreements; it is coordinating those agreements for more effective collective middle-power leverage.

Comparable challenges arise in security and digital governance. When major geopolitical crises occur, no standing forum exists where middle powers can confer privately and determine whether coordinated global responses would better advance their interests and support the international rules-based order. Likewise, issues such as artificial intelligence, cybersecurity and digital platform governance increasingly place the sovereignty and autonomy of individual middle powers opposite to multinational technology companies possessing immense economic and political influence. Essentially, acting collectively would significantly strengthen middle powers’ capacity to respond.

Existing initiatives, such as MIKTA and the Alliance for Multilateralism, demonstrate both the possibilities and limitations of informal modes of cooperation. Both have facilitated dialogue, but, lacking permanent institutional structures, they have struggled to sustain broader strategic influence.

The Organisation for Economic Co-operation and Development (OECD) offers a more instructive model. Much of its effectiveness derives not simply from its membership but also from its permanent secretariat, its significant analytical and operational capacity, and robust deliberative processes capable of developing both legal and soft-law instruments across a wide range of issues. These institutional characteristics have enabled the OECD to exert influence well beyond its formal authority. A middle-power institution should draw upon these strengths while avoiding governance arrangements that permit domination by any single major power.

The broader lesson is straightforward: permanent institutions reduce transaction costs by providing continuity, institutional memory, trusted deliberative processes and analytical support. Rather than requiring governments to recreate relationships and organizational mechanisms whenever new issues arise, they create the conditions under which cooperation becomes more timely, efficient and effective.

Variable geometry should therefore not be viewed as an alternative to institutions. Flexible coalitions work best when supported by a stable institutional framework that enables them to form, evolve and dissolve as circumstances require.

A Middle-Power Room of Their Own

The proposal advanced here is modest but potentially transformative. Middle powers should establish a permanent institutional home through which they can deliberate collectively while retaining complete autonomy in their external relationships.

Such an institution would neither replace the United Nations nor compete with existing regional organizations. Nor would it constitute a geopolitical bloc. Participation would not require common positions on every issue, nor would it restrict bilateral or multilateral relationships with great powers. Rather, it would provide a standing forum, permanent analytical capacity and organizational continuity through which middle powers could determine when (and with whom) it makes sense to cooperate.

As a practical first step, a representative founding secretariat should be established to draft proposals on membership, governance, funding and organizational structure. These proposals would then be finalized at an international conference. Establishing joint headquarters in both the Global North and the Global South would clearly signal a commitment to the realities of global governance in the twenty-first century.

To be sure, significant operational challenges lie ahead. These include addressing definitional hurdles regarding what constitutes a “middle power”; aligning on a shared value system, such as the rule of law, human rights, free markets and institutional accountability; and establishing flexible, fit-for-purpose, deliberative decision-making processes. While responding effectively to challenges such as these will undoubtedly require ingenuity, history demonstrates that diverse groupings of countries — when united by a common purpose — can overcome these obstacles while respecting individual differences.

Governance evolves because the world evolves. The institutional architecture established after the Second World War reflected the geopolitical realities of its time. Today’s multi-polar world is characterized by a broader distribution of influence in which middle powers have become indispensable contributors to trade, technology, security and the maintenance of the international rules-based order. Recognizing that reality through a permanent institutional home for middle powers would improve their collective leverage but would not diminish existing institutions. It would strengthen the global governance architecture by filling one of its most significant institutional gaps.

The twentieth century required institutions designed primarily around the great powers. The twenty-first century needs the development of an institution reflecting the increasingly consequential role played by middle powers that leverages their interests and ideas for maximum influence on the global stage. It is time that middle powers had a room of their own.

The opinions expressed in this article/multimedia are those of the author(s) and do not necessarily reflect the views of CIGI or its Board of Directors.

This opinion’s analysis was developed and approved by the named human authors under documented review controls; AI assistance was judicious, disclosed and verified. Learn more about CIGI’s Human Analysis Standard.

About the Author

Kernaghan Webb is an associate professor in Toronto Metropolitan University’s Department of Law and Business. He is also the Founding Director of the Toronto Metropolitan University Institute for the Study of Corporate Social Responsibility.

For more insights, visit CIGI Online.

What was the main message of Prime Minister Mark Carney’s now-famous speech in Davos earlier this year?

There are no fewer than three competing takeaways from Carney’s address — and a recent op-ed for The Economist co-authored with Finnish President Alexander Stubb has done little to reconcile them. Now, more than half a year after Carney’s speech and for all the focus on the current inflection point in the international order, there is little clarity on what the Carney government’s vision for a revised and more resilient order might look like. This is possibly because, despite considerable global turbulence, it is Canada’s relationship with its southern neighbour rather than the international order itself that has reached a breaking point.

The central theme in Carney’s address was his call to “take down the sign.” For years, Western countries claimed to act in defence of the “rules-based international order” when this was never verily a neutral description of the postwar world. In reality, appeals to the “rules-based order” were a discourse born out of anxiety over Western decline, aiming to write the rules of the future by shoring up the power structures of the past. Nor did it help that the most vocal defenders of this supposed order were prepared to tolerate a highly selective application of the rules.

But now that we have taken down the sign, what comes next?

The first possibility hinted at is more realism and less megaphone diplomacy. As we are constantly told, we must take the world as it is rather than as we wish it to be (as if a world in which non-Western actors have power and agency is somehow a lamentable fact).

Carney and Stubb rightly note that progress “can be signalled through monologue, but it will be achieved only through dialogue.” After Ottawa’s ties with Beijing and New Delhi rapidly declined under Justin Trudeau’s premiership, this is a message that Canadians need to hear. One wonders, of course, whether the Carney government will be committed to privileging dialogue over monologue when it comes to Russia, whose exclusion from the core of the post-Cold War European security order engendered a security dilemma that resulted in war. Still, employing the vocabulary of interests is a much welcome respite to erstwhile lectures about “who we are” that were of little interest to the rest of the world.

Subscribe to our newsletter

The second — and more idealistic — option that Carney called for in Davos involves “acting consistently, applying the same standards to allies and rivals.” This is a test that Carney appeared to fail after the United States launched a war of choice against Iran in violation of the United Nations Charter. It was one that he continued to shirk at the very outset of his op-ed with Stubb, where Russia is accused of “waging a war of aggression against Ukraine and actively interfering in other societies” while the United States is merely characterized as “reassessing its priorities and transforming all its commercial relationships” despite Donald Trump’s repeated threats to annex Canada and Greenland.

Between cynical realpolitik and a consistent commitment to international norms lies a third path — building a coalition of middle powers based on “values-based realism” and principled pragmatism. This approach views ad hoc minilateral partnerships — or “variable geometry” — as an alternative to what Carney in Davos called “naïve” multilateralism on the one hand and multipolarity, which Carney and Stubb associate with “chaos,” on the other.

Such calls for a via media are not particularly novel. Already a decade ago, then Foreign Affairs Minister Stéphane Dion was advocating for a foreign policy based on “responsible conviction,” which sounds an awful lot like Carney’s call to be both “principled and pragmatic.”

Leading Global South actors such as India and Brazil have long been pursuing multi-vectored foreign policies. It is rather the West that is beginning to emulate a trend already set by others — yet Carney and Stubb conclude their op-ed by inviting others to join them. It was not all that long ago, during Chrystia Freeland’s tenure at the top echelons of government, that cooperation with “likeminded” countries was the talk of the town. Carney and Stubb concede that “not every partner will share all of our values” but their proposed remedies centre mostly on working with the same group of friends: the North Atlantic Treaty Organization, the European Union and the Group of Seven.

Whether a global coalition of middle powers is even possible at a time when great powers have demonstrated their willingness to wield considerable leverage to keep wayward states in line remains an open question. A more “variable” configuration may prove more feasible than a cohesive coalition. But the more germane question is what problem do such partnerships purport to solve? If the purpose is greater capacity for self-help and strategic autonomy in a Machiavellian world, then this may be desirable. But this is far from a panacea for the supposed “rupture” in the international order that we are told we are living through.

Carney and Stubb do not mince words: they claim that the “world order built after 1945 is unravelling.” But a crisis of authority within an international order is not the same as a crisis of the order itself. The so-called rupture at which we have arrived is the product of decades of contestation over the rules and who gets to write them. Unless we address the sources of these disagreements, we will remain on a pathway toward the law of the jungle that ad hoc coalitions alone may not be able to prevent.

The post-1945 order has gone through several phases, from hot rivalry to détente and from neoliberal globalization to renewed great power competition. Throughout this time, the legal and institutional architecture of the order has survived and even grown denser. Rules may be bent or broken but no one fundamentally challenges the legitimacy of international law and multilateralism as core organizing principles for interstate relations.

Canada’s relationship with the United States may be undergoing a rupture but this is not necessarily so for the international order. This is because American hegemony was never synonymous with the postwar order writ large — a revisionist belief that the unipolar moment of the 1990s helped to foster. Rather, it was a feature of the post-1991 order based on a revived standard of civilization that liberal states sought to substitute for the post-1945 order.

While it may be a useful message for domestic audiences that have grown unaccustomed to the necessity of strategic choices and trade-offs, being both principled and pragmatic is a formula for justifying almost any policy. We have not yet reached a complete rupture in the international order but we may yet unless we seize what Carney’s co-author calls our “last chance” to treat rival powers and the global majority as sovereign and equal rule-makers.

The opinions expressed in this article/multimedia are those of the author(s) and do not necessarily reflect the views of CIGI or its Board of Directors.

This opinion’s analysis was developed and approved by the named human authors under documented review controls; AI assistance was judicious, disclosed and verified. Learn more about CIGI’s Human Analysis Standard.

About the Author

Dr. Zachary Paikin is deputy director of the Better Order Project and research fellow in the Grand Strategy Program at the Quincy Institute for Responsible Statecraft. He is also senior fellow at the Institute for Peace and Diplomacy.

For more analysis, visit CIGI Online.

For years, debates about autonomous weapons and artificial intelligence (AI)-assisted targeting remained largely theoretical. Critics warned of future harms, while some powerful countries insisted existing legal frameworks and human oversight would ensure responsible use. The absence of identifiable victims allowed many to treat these concerns as speculative.

That changed this year. Following a US strike on a girls’ school in southern Iran that killed at least 157 people, most of them students, public attention quickly turned to whether AI-assisted targeting had played a role. There was a strong indication that AI targeting was being used in the war, with some 1,000 targets struck in the first 24 hours. Although subsequent reporting suggests outdated intelligence — rather than AI itself — was likely responsible, the debate exposed deeper concerns about the quality of intelligence feeding military decisions, the pressure to generate targets rapidly and the growing reliance on AI-enabled systems operating on imperfect data. These questions became even more urgent following a joint investigation by The Independent and Airwars, which identified Abdul-Rahman al-Rawi, a 20-year-old student killed in a US strike in Iraq in 2024, as the first publicly known victim of an airstrike in which AI-assisted targeting was officially acknowledged.

For most people far from conflict zones, autonomous weapons and AI-enabled targeting may still seem abstract. However, the issue matters because the rules governing war shape how violence is used — and misused — against real people. Those rules exist precisely to constrain violence and protect civilians, and they are under pressure.

The question governments should now be asking is how AI is changing human judgment in war. As these systems become embedded across military planning and targeting, understanding how they shape what commanders see, how quickly they act, and how they exercise legal and ethical judgment are just as important as understanding the technology itself.

Military organizations are expected to seek decision advantage. AI systems promise to process vast quantities of information, identify patterns that humans might miss, generate target recommendations and accelerate what militaries call the “targeting cycle.” In increasingly complex operational environments, these capabilities appear to offer clear military benefits. Proponents often argue that AI tools may improve situational awareness, reduce cognitive burden and enable faster responses to evolving threats. But military advantage, or the promises of the technology, should not be confused with better decision making or the practical realities.

Subscribe to our newsletter

Judgment in war has never been simply about making decisions quickly. It requires weighing uncertainty, interpreting context, questioning assumptions, and balancing military necessity against legal and humanitarian obligations. AI can support those processes, but it also changes the conditions under which they occur. When AI systems produce recommendations at a pace that humans struggle to interrogate, or when operational success becomes associated with the speed and scale of targeting, the opportunity for meaningful deliberation begins to shrink. As a result, human judgment risks becoming increasingly procedural rather than genuinely reflective. This matters because the consequences extend far beyond individual strikes.

When targeting decisions are made faster, mistakes can also spread faster. Civilians are often the first to bear those consequences, particularly in densely populated urban environments where distinguishing combatants from civilians is already difficult.

For more than a decade, governments have debated these challenges through discussions on lethal autonomous weapons systems at the United Nations. A broad consensus has emerged that existing international humanitarian law applies to AI-enabled weapons and that humans must retain responsibility for decisions to use force. Yet recent conflicts suggest that affirming these principles is becoming easier than implementing them. “Meaningful human control” has become one of the defining concepts in these debates, more recently expanded to include context-appropriate human judgment and control. But invoking these concepts is not enough. The more difficult question is whether meaningful human judgment is actually possible under the operational conditions that AI is helping to create. Do commanders understand how AI-generated recommendations are produced? Can operators realistically question the outputs of increasingly complex systems while working under intense time pressure? Does the technology gradually narrow the range of options considered acceptable, or nudge certain decisions over others through system design?

These questions deserve far greater attention because AI does more than generate recommendations. As scholars and analysts highlight, it increasingly shapes the decision environment itself, what information is prioritized, which options appear most credible and how much time decision makers have to exercise independent judgment. Even when a human formally authorizes a strike, the quality of that judgment may already have been influenced by the design of the system, the data it relies upon and the pace at which decisions are expected to be made.

This is why governance must extend beyond the moment a weapon is employed. Accountability begins long before a commander authorizes the use of force. It begins with the data selected to train AI systems, the assumptions built into algorithms, the testing and validation processes used before deployment, and the institutional incentives that reward speed and efficiency. It continues through operational use and must include rigorous post-strike review and mechanisms for learning when systems fail.

States therefore have opportunities to strengthen governance now, even if negotiating new international agreements remains politically difficult. Key is to move beyond treating human judgment as a procedural requirement and instead ensure it remains a meaningful operational capability. That means providing commanders and operators with sufficient training, time, authority and institutional support to critically evaluate AI-generated recommendations rather than simply approve them.

Additionally, governments should significantly improve transparency around AI-enabled military systems. While sensitive operational information will always require protection, states should be clearer about how these systems are tested, validated, monitored and evaluated against their legal obligations. Without greater transparency, meaningful accountability becomes exceedingly difficult.

Finally, states should strengthen oversight and accountability mechanisms across the entire life cycle of AI-enabled military systems. This should include documentation of decisions made throughout the life cycle by the operators, commanders and legal advisers involved, in a form that enables genuine traceability of accountability and supports independent review, not just internal record keeping. Such documentation would also encourage a greater understanding of the role of human judgment and how it was, or was not, enabled in practice.

Abdul-Rahman al-Rawi was not a theoretical example. He was a student whose death has given a human face to debates that governments have conducted largely in the abstract for more than a decade. In the coming months, states will decide on the future of the discussions on autonomous weapons as well as ways to move forward on broader military AI dialogue. The question before them is how to ensure that human decision makers retain the capacity to exercise careful, informed and independent judgment in environments increasingly shaped by AI. Protecting civilians in the age of AI will ultimately depend on the answer.

The opinions expressed in this article/multimedia are those of the author(s) and do not necessarily reflect the views of CIGI or its Board of Directors.

This opinion’s analysis was developed and approved by the named human authors under documented review controls; AI assistance was judicious, disclosed and verified. Learn more about CIGI’s Human Analysis Standard.

About the Author

Branka Marijan is a CIGI senior fellow and a senior researcher at Project Ploughshares.

This article was first published by Canada’s National Observer. For more analysis, visit CIGI Online.

Recent global events, from war-driven oil shocks to attacks on critical infrastructure, have shown the fragility of current centralized fossil fuel chains. The effects have been felt most prominently at the pump, resulting in an average fuel price increase of 27.1 per cent since the closing of the Strait of Hormuz. As Canada seeks more resilience from these shocks, shown by the recently published national electricity strategy and broader electrification agenda, no product highlights this more than electric vehicles (EVs). Right now, the more important policy question is no longer whether EVs are cleaner or more efficient, but which transportation and energy systems are likely to prove more resilient in an era defined by geopolitical fragmentation, cyber risk, climate disruption and information disorder.

With that in mind, the more important distinction may be between centralized fossil fuel systems and increasingly distributed electric systems. Gasoline and diesel supply chains depend on highly centralized infrastructures of extraction, refining, transport and retail distribution. These systems are vulnerable to geopolitical conflict, cyberattacks, natural disasters, shipping disruptions and price volatility. Recent global conflicts and energy crises have repeatedly demonstrated how quickly fuel prices and supply chains can destabilize when concentrated infrastructure is disrupted. Even countries like Canada and the US — two of the top oil producing countries in the world — are not immune from such commodity shocks and subject to international oil prices.

For years, EV critics have framed them as fragile, overly digital machines: vulnerable to cyberattacks, dependent on unreliable infrastructure, and incapable of functioning during serious crises. In public discourse, gasoline-powered internal combustion engine (ICE) vehicles are often portrayed as the more rugged and resilient option — particularly during geopolitical instability, infrastructure disruption or emergencies. But many critiques directed specifically at EVs misunderstand a broader reality: modern vehicles of all kinds are now deeply digital systems.

Regardless of perceptions, today’s ICE vehicles are also “computers on wheels.” Modern automobiles — regardless of powertrain — rely heavily on software controlling braking, steering, navigation, safety systems, engine management, connectivity and remote updates. Many contain more software code than advanced military aircraft. As vehicles become increasingly connected and autonomous, cybersecurity risks are becoming systemic across the automotive sector and not unique to EVs. The relevant comparison is therefore not between “digital EVs” and “mechanical gas cars.” That world no longer exists.

While it is true that electricity systems are not immune to disruption, they can also draw from a far wider range of generation sources and distribution models. Power can increasingly be produced domestically through hydroelectricity (approximately 55 per cent of Canada’s electricity generation), nuclear energy (13 per cent), wind (7 to 8 per cent), solar (1 per cent), geothermal and distributed microgrids, reducing reliance on vulnerable centralized energy systems. In countries such as Canada, where electricity generation is already relatively diversified and low-carbon, this creates important long-term resilience advantages by reducing exposure to global fuel price volatility, strengthening energy security, increasing system redundancy, improving adaptive capacity in the face of climate-related disruptions and enabling more decentralized and flexible energy systems capable of sustaining critical infrastructure during emergencies.

That said, EVs are vulnerable to a different set of problems, such as challenges in battery supply chain dependencies, charging infrastructure gaps, fragmented provincial policy frameworks and cybersecurity concerns. Canada and its allies also remain heavily dependent on critical minerals and battery manufacturing capacity, of which China controls between 70 and 96 per cent, along with other geopolitically sensitive supply chains.

EV charging infrastructure is also becoming more distributed and adaptable over time. While charging networks still require substantial expansion, EVs can already be charged from homes, workplaces, public chargers, solar installations, backup battery systems and localized grids. During emergencies, access to electricity may ultimately prove easier to restore than liquid fuel logistics, dependent on tanker deliveries, refinery operations and functioning gas stations. Even gasoline infrastructure itself relies heavily on electricity for fuel pumps, payment systems, refining operations and transportation networks. Concerns about EV battery fires are often poorly contextualized, as while lithium-ion fires can be difficult to extinguish, gasoline vehicles catch fire at significantly higher rates, with EV incidents receiving disproportionate attention due to their novelty and media visibility.

At the same time, public debate surrounding EVs has become increasingly shaped by misinformation and disinformation circulating across digital platforms. Viral claims about inevitable grid collapse, exaggerated battery fire risks or the impossibility of EV adoption in cold climates frequently spread without context, often amplified through polarized online ecosystems. Legitimate debates certainly exist around infrastructure readiness, industrial policy, mineral supply chains and affordability. These include questions about the expansion of EV charging networks and electricity grids, the development of domestic clean technology and manufacturing capacity, reliance on critical minerals such as lithium, cobalt and nickel, and the costs associated with electric vehicles and other low-carbon technologies. But distorted or misleading narratives can undermine evidence-based policymaking and complicate long-term infrastructure planning precisely when governments are attempting to strengthen energy resilience.

This creates an important policy challenge. Governments should not simply promote EV adoption; they should also strengthen public understanding of the broader resilience benefits associated with electrification. Public-interest communication strategies, greater transparency around energy infrastructure planning, and support for independent research on online energy misinformation could all help improve the quality of democratic debate surrounding transportation transition policies.

Policymakers should also work with provinces, utilities and industry to accelerate investment in distributed grids, vehicle-to-grid technologies, interoperable charging networks and domestic battery recycling and manufacturing capacity. Strengthening cybersecurity standards for all connected vehicles and charging systems will likewise become increasingly important as transportation and energy systems converge.

None of this suggests an immediate or universal transition away from ICE vehicles. In remote regions and sectors requiring specialized heavy-duty transportation, conventional and hybrid vehicles may remain necessary for years to come. Nor should policymakers underestimate the importance of securing critical mineral supply chains and addressing vulnerabilities within electricity infrastructure itself.

When centralized fuel chains fracture, the costs land on ordinary Canadians first and fastest. EVs won’t eliminate that vulnerability overnight, but they point toward a system that is distributed, domestically grounded and harder to shock. In an era defined by energy insecurity, digital dependence, geopolitical volatility and increasingly contested information environments, resilience increasingly depends on diversification, decentralization and adaptability. Electrified transportation systems, despite their limitations, may ultimately align more closely with those resilience imperatives. Canada’s national electricity strategy is a start — the harder work is building the governance institutions to match the ambition.

The opinions expressed in this article/multimedia are those of the author(s) and do not necessarily reflect the views of CIGI or its Board of Directors.

This opinion’s analysis was developed and approved by the named human authors under documented review controls; AI assistance was judicious, disclosed and verified. Learn more about CIGI’s Human Analysis Standard.

About the Author

Andrew Heffernan is a CIGI fellow. He holds a Ph.D. in political science from the University of Ottawa, where he is an adjunct professor teaching on global governance, environmental politics and information integrity.

For more analysis, visit CIGI Online.

Two developments at the recent World Artificial Intelligence Conference (WAIC) in Shanghai sent the same signal. China launched the World Artificial Intelligence Conference Organization (WAICO) to institutionalize its vision of more open and inclusive global AI governance, while Moonshot AI unveiled Kimi K3, China’s most powerful open-weight model to date. Together, they suggest that China is likely to remain committed to open-source and open-weight AI — for now.

WAICO institutionalizes China’s preferred AI governance message: AI should be accessible to all countries, development should be prioritized alongside safety, and global coordination should remain closely linked to the United Nations. While President Xi Jinping’s WAIC address, the 2026 WAIC chair’s statement and the announcement establishing WAICO did not explicitly commit China to open-source AI models, they emphasized openness, inclusion and broader access to AI, especially for developing countries.

This approach reflects China’s broader AI development strategy. Open models allow Chinese companies to attract developers, accelerate industrial applications and expand overseas adoption, especially in countries that lack advanced AI capabilities. More importantly, they enable Beijing to internationalize its domestic AI strategy by extending China’s open-source ecosystem beyond its borders. Through application-driven innovation, “AI Plus” initiatives and AI-enabled industrial upgrading, China is seeking to build an ecosystem around its own technologies while promoting AI as an international public good rather than an exclusive technology controlled by a handful of countries. In doing so, Beijing advances its longer-term objective of expanding global AI infrastructure and increasing the international reach of Chinese AI.

Rather than serving simply as another international AI forum, WAICO institutionalizes China’s strategy by providing China with a permanent platform to shape global AI governance. Drawing on China’s previous experience creating international organizations, such as the Asian Infrastructure Investment Bank and the New Development Bank, WAICO appears designed to combine institution building with practical cooperation. China is likely to provide much of the financial support for capacity building, technical assistance and AI infrastructure projects in participating countries, particularly across the Global South.

This open-source diplomacy, however, is unlikely to persuade established Western powers, which remain committed to smaller, values-based AI governance frameworks. The geopolitical alignment is difficult to miss. In this respect, China’s message of openness and public goods resembles Xi’s earlier defence of free trade at Davos days before Donald Trump’s inauguration in 2017: rhetorically attractive, but received in the West with muted responses and considerable skepticism.

Subscribe to our newsletter

Compared with initiatives such as the Global Partnership on AI (GPAI) and the Council of Europe Convention on AI, WAICO stands out in three respects: its emphasis on AI capacity building for developing countries, its promotion of openness and open-source ecosystems, and its close alignment with the United Nations. Together, these features reveal China’s broader geopolitical ambitions. By advocating openness, opposing what it describes as the over-securitization of AI, and promoting a UN-centred governance framework, Beijing seeks to position itself as a champion of equitable access to AI while implicitly criticizing US-led export controls and technology restrictions. Rather than simply participating in existing AI governance frameworks, WAICO advances a competing vision — one that prioritizes technology diffusion, development and the interests of the Global South.

WAICO’s membership reinforces this geopolitical reading. Unsurprisingly, its 29 initial signatories are exclusively non-Western states, spanning Russia, Brazil, Indonesia and Pakistan alongside other members of BRICS, the Shanghai Cooperation Organization, as well as 10 African countries. The absence of Western members suggests that WAICO may become not so much a universal AI governance platform as a China-centred coalition focused on development, infrastructure and resistance to Western technology controls.

WAICO’s near-term priorities translate this broad rhetoric into a more concrete institutional strategy. The organization will focus on capacity building — including Xi’s pledge to provide 5,000 training opportunities for developing countries and to establish AI cooperation centres with the Association of Southeast Asian Nations, the African Union, BRICS and other regional bodies — while supporting an open-source ecosystem designed to extend China-centred AI infrastructure abroad and anchoring its governance agenda in UN-centred multilateralism. Together, these priorities reflect China’s view that development deserves as much attention as safety, a position it also advanced through the AI + International Cooperation Initiative proposed at the UN Global Development Initiative high-level meeting in September 2025.

This UN alignment is already visible. With the UN Secretary-General António Guterres attending and endorsing the event, WAICO is likely to promote goals that overlap with the UN agenda, including AI for good, human-centred AI. The alignment was also reflected in the China-backed UN resolution on strengthening international cooperation in AI capacity building at the 78th UN General Assembly in 2024, as well as in China’s proposed “AI Capacity-Building Action Plan for Good and for All” released in September 2024. Both prioritize global AI connectivity and digital infrastructure, especially in the Global South, and the application of AI to empower various industries.

Open Weights as Business Strategy

Yet institutional initiatives alone do not explain why China continues to embrace openness. Equally revealing are the decisions made by China’s leading AI companies. The release of Kimi K3 and DeepSeek’s continued commitment to open-source development provide perhaps the clearest evidence that openness remains a strategic choice for China’s AI industry — not merely a political slogan. For now, China appears likely to maintain both the concept and practice of open source, but the durability of this commitment depends on whether Chinese firms continue to see openness as strategically useful once the technology gap narrows.

Kimi K3, released during WAIC, is one of China’s most powerful AI models and further narrows the gap with leading US models such as Claude Fable 5 and GPT 5.6 Sol. Despite the exaggerated claims that Chinese AI models are overtaking their US counterparts, Kimi has acknowledged that K3 still lags behind leading American models. More importantly, Kimi announced that K3 would remain open weight.

DeepSeek appears to be following the same path. According to the recent comments by founder Liang Wenfeng, the company intends to continue pursuing open-source development, potentially including its most advanced models. Liang argues that openness does not weaken DeepSeek’s competitive position because releasing model weights does not eliminate the formidable technical and financial barriers to deploying frontier AI systems. The barrier to entry remains extremely high even when a model is open source: deploying it effectively is difficult and running it at very low cost is even harder. After all, deploying frontier AI systems effectively and at low cost requires infrastructure, engineering and computing resources that cannot simply be copied by downloading an open model.

Liang’s view provides an important corporate-level counterpart to China’s official rhetoric. Open source helps Chinese firms expand adoption, strengthen ecosystems and compete with leading US companies without necessarily sacrificing commercial returns. Liang argues that open source and commercialization are not mutually exclusive: DeepSeek can support competitors and community deployment while still preserving advantages in cost control, engineering and operational efficiency. The AI market is sufficiently large that no single company can dominate it, making ecosystem expansion more valuable than maximizing short-term licensing revenue.

For now, open source serves China’s strategic interest on every front. While China is still catching up, openness helps diffuse Chinese models, build ecosystems, reduce dependence on Western platforms and gain support from developing countries. But if Chinese firms further close the gap or achieve leadership, they may have stronger incentives to protect model weights, talent and intellectual property.

Yet China’s openness has limits. Beijing is also working to protect AI-related intellectual property, talent and technologies. Reports of efforts to prevent talent from leaving the country, the forced cancellation of Meta’s proposed $2 billion acquisition of Manus, and Financial Times’ reporting on export controls restricting foreign access to China’s most advanced AI model weights all point in the same direction. If China achieves, or comes close to achieving, AI technological leadership, the incentives may shift from diffusion to protection.

The opinions expressed in this article/multimedia are those of the author(s) and do not necessarily reflect the views of CIGI or its Board of Directors.

This opinion’s analysis was developed and approved by the named human authors under documented review controls; AI assistance was judicious, disclosed and verified. Learn more about CIGI’s Human Analysis Standard.

About the Author

Alex He is a CIGI senior fellow. Alex is an expert on digital governance in China, the Group of Twenty (G20), China and global economic governance, domestic politics in China and their role in China’s foreign economic policy making, and Canada-China economic relations.

For more insights, visit CIGI Online.

In 2021, global supply chains buckled under the combined pressures of the COVID-19 pandemic and widespread logistical disruptions. A year later, in 2022, the European Union woke up to its dangerous dependence on Russian energy. In 2026, the next strategic vulnerability staring us in the face is workforce dependence. Most nations maintain sophisticated systems to detect military threats, economic shocks and logistical breakdowns. By contrast, although several countries now conduct sophisticated workforce and skills anticipation, few integrate those efforts across government into a strategic warning capability that informs national priorities.

While governments and corporations devote enormous resources to understanding risks to capital, trade, energy and security, human capital — widely recognized by institutions such as the World Bank and the Organisation for Economic Co-operation and Development (OECD) as a fundamental driver of long-term productivity, resilience and national prosperity — remains managed largely through backward-looking statistics. Labour-force participation, vacancies, unemployment and retirement projections serve an important administrative purpose, but they simply describe conditions that already exist rather than emerging vulnerabilities.

As countries compete for technological leadership, industrial resilience, health-care capacity, Arctic development and strategic autonomy, the ability to anticipate workforce constraints may prove just as important as anticipating geopolitical ones. The next great competition among advanced economies may not be over resources, markets or even technology. It may be over the human capabilities required to transform ambition into real capability.

Building on the literature on labour-market intelligence, strategic foresight and state capacity, “workforce intelligence” can be understood as the systematic identification of emerging human-capital risks before they evolve into economic, industrial or strategic constraints.

The purpose of workforce intelligence is to provide early warning about future capability gaps. Unlike workforce planning, which aligns staffing requirements, or labour-market forecasting, which estimates supply and demand, workforce intelligence synthesizes demographic, industrial, educational, immigration and geopolitical signals to assess whether future workforce conditions could constrain national objectives. Essentially, while labour-market analysis asks What is happening?, workforce intelligence would ask, What will happen and what does it mean for national power?

Human capital occupies a unique position within national power. Military capability, economic productivity, technological innovation and diplomatic influence ultimately depend on people with the skills to sustain them. Canada can sit on vast critical mineral deposits, but without enough mining engineers, geologists and skilled tradespeople, those resources will stay in the ground. It can increase defence spending, but without shipbuilders, technicians and systems engineers, the platforms never materialize at scale. It can launch ambitious artificial intelligence strategies and innovation funds, but without sustained access to elite technical talent, investment risks generating visibility rather than capability.

Subscribe to our newsletter

Political scientists have long argued that states succeed not merely because they possess resources or policy vision, but because they possess state capacity — the ability to translate objectives into outcomes. Workforce intelligence should be understood as a neglected component of state capacity.

A practical workforce intelligence framework would rest on five domains: demographic intelligence, to anticipate retirements; talent intelligence, to track globally scarce skills and migration; educational intelligence, to align graduate supply with strategic priorities; industrial intelligence, to forecast workforce needs for major national projects; and geopolitical workforce intelligence, to assess how strategic competition shapes the global movement of talent.

Canada’s Fragmented Picture

Canada’s strategic agenda increasingly centres on economic resilience and strategic autonomy through critical minerals, Arctic development, defence modernization, infrastructure renewal and advanced technology. Less glamorous is the fact that each of those depends on workforce capacity that current systems cannot guarantee. Planning remains fragmented across provinces, federal departments, educational institutions and private employers, with no single entity holding the full picture. Yet strategic autonomy ultimately depends on capability, and capability depends on people.

Canada maintains a national security apparatus and sophisticated economic policy institutions. What remains absent is an integrated workforce intelligence capability that translates workforce information into strategic warning and national capability assessments. Canada already has important labour-market institutions. Employment and Social Development Canada (ESDC) and Job Bank provide data, occupational outlooks, wages and employment trends, while the Future Skills Centre supports research and innovation in skills development. Canada also created the Labour Market Information Council in 2017 to improve the sharing of labour market information across jurisdictions. Its mandate, however, is the quality and accessibility of the information itself, not assessment of what workforce conditions mean for national objectives. These functions are valuable, but they are not the same as workforce intelligence. Their primary focus is information, skills development and workforce adaptation, rather than how those signals combine into forward-looking assessments of national capability.

Intelligence organizations exist to identify emerging vulnerabilities before they become operational failures. They ask: What is changing and what are the implications? Workforce policy, by contrast, usually reacts after the fact — often too late and at higher cost.

The distinction matters. A vacancy rate may show that a shortage exists, while workforce intelligence would ask whether that shortage threatens a defence procurement timeline, a critical minerals project, an Arctic infrastructure plan or Canada’s ability to retain strategic technology talent. In other words, its value lies not in producing more labour-market data, but in interpreting workforce signals through the lens of national objectives.

The next step for Canada would be to establish a national workforce intelligence function, whether housed within an existing central agency or organized as a dedicated analytical centre. Its role would not be to duplicate ESDC, Statistics Canada, Job Bank or the Future Skills Centre; instead, it would synthesize their outputs with industrial policy, immigration trends, education pipelines and strategic priorities. The result would be a forward-looking capability assessment to identify where Canada’s ambitions are likely to collide with human-capital constraints.

Other countries are moving in this direction, though none offers a perfect model. Jobs and Skills Australia produces employment projections and workforce analysis to inform policy and planning. Singapore’s SkillsFuture publishes its annual Skills Demand for the Future Economy reports to identify emerging skills needs in areas such as the digital, green and care economies. The OECD has similarly emphasized whole-of-government skills governance, integrated information systems and stronger coordination across departments. These examples show that advanced economies are beginning to treat skills and workforce capacity as strategic infrastructure. Canada should adapt that lesson but extend it further: from labour-market planning to national capability assessment.

Major federal initiatives should undergo workforce risk assessments alongside financial and environmental reviews. A critical minerals strategy, defence procurement plan or Arctic infrastructure project should identify not only its budget and environmental risks, but also the occupations, regions, education pipelines and immigration pathways on which delivery depends. Canada should also publish an annual workforce intelligence review focused on capability gaps that could constrain national objectives.

Ultimately, the defining strategic failures of the twenty-first century may not arise from a lack of capital, resources or political ambition. They may arise from an inability to anticipate workforce vulnerabilities before they become capability failures. Governments have spent decades developing institutions to monitor military threats, financial instability and geopolitical risk. The next challenge is building equivalent capacity to understand human-capital risk with the same seriousness.

The opinions expressed in this article/multimedia are those of the author(s) and do not necessarily reflect the views of CIGI or its Board of Directors.

This opinion’s analysis was developed and approved by the named human authors under documented review controls; AI assistance was judicious, disclosed and verified. Learn more about CIGI’s Human Analysis Standard.

About the Author

Abbas Qaidari is a researcher specializing in international security and Arctic defence strategy.

For more analysis, visit CIGI Online.

The people building the most advanced artificial intelligence (AI) seem to think the race is about intelligence: creating the smartest, most capable AI systems. The leaders of Anthropic, DeepMind and OpenAI regularly posture about the imminent arrival of artificial general intelligence. When their companies promote new models, the first few paragraphs always highlight capabilities: the reasoning score, coding result or gap in performance over the last release. Other factors, such as cost or privacy, are buried after long descriptions of the breakthrough intelligence of new systems. The assumption that underlies this signalling is that the winner will be whoever builds the smartest model.

This is a mistaken premise. The developer who wins the AI race will not be the one with the most capable model, but rather the one who can deliver good-enough intelligence at the lowest sustainable price. There are naturally different definitions of what winning means, such as achieving the greatest market share or producing the highest-quality technology. I understand winning in economic terms, as Anthropic and OpenAI are, after all, for-profit businesses. Essentially, victory means achieving the greatest degree of durable diffusion: producing AI systems that are embedded in global economic processes and infrastructure.

History tells us that being first or being the best does not guarantee winning. Andrew Carnegie, for example, did not become the steel giant of the nineteenth century by first discovering or refining the purest steel. He won because, through vertical integration, ruthless cost accounting and the embrace of new technological processes, he produced high-quality steel more cheaply than any of his competitors. The mid-twentieth-century semiconductor boom teaches similar lessons. In the late 1960s, DRAM (dynamic random-access memory), an important semiconductor memory architecture, was first created in the United States. By the 1980s, the Japanese led in market share due to their high-quality and low-cost production methods.

The lesson here echoes today: The country that creates the most capable AI systems — also known as frontier — is not guaranteed to own what the frontier becomes once it is cheap. This is arguably already occurring with AI. The transformer — the architecture that underpins virtually every major modern large language model — was discovered by Google, an American company, in 2017. Today, Chinese labs such as DeepSeek and Alibaba have shown that they can consistently develop transformer-based models at a fraction of the cost of their Western competitors. By my own calculations, as of May 2026, DeepSeek’s best model was only three percent worse in quality than the best American model but 96% cheaper.

Subscribe to our newsletter

The Rising Cost of AI

Cost leadership in AI is unlikely to come from race-to-the-bottom price wars in which companies routinely undercut one another with unsustainably low prices in order to capture greater market share. Durable cost leadership will come from structural business advantages, such as owning the supply chain, the hardware or energy. As such, pure-play laboratories that are burning massive amounts of cash to stay a few points ahead on a benchmark are the most exposed.

In fact, the cost problem has already arrived. This past spring, an Nvidia executive noted that the cost of tokens for his team already exceeded the cost of the employees they were meant to augment. Uber’s COO likewise admitted that the company’s entire 2026 budget for AI-coding tools had been exhausted by April. Another enterprise, according to Axios, spent half-a-billion dollars on Claude in a single month after failing to cap employee usage. In other words, cost is becoming the main AI business story of 2026. Moving forward, the question about AI will be less about what a model can do, and more about its costs and if any developer can actually build it sustainably.

Chinese developers, such as DeepSeek and Alibaba, have come to terms with this intuition first. They now offer models nearly as capable as their Western rivals at a fraction (more than a tenth) of the price. These pricing discounts are likely authentic and genuinely structural. Many of the best Chinese models are open weight, backed by the state, powered by cheap energy and made by companies that tolerate thin margins. These structural realities have important governance implications. Western export controls, which have been a pillar of US-led AI policy against China, are designed to prevent China from creating frontier AI models. However, if the AI race is decided by diffusion, which, in turn, hinges on price, controls aimed at stalling advances in raw capabilities will do little to prevent Chinese dominance in the cost layer.

There is already evidence that DeepSeek is becoming one of the world’s most widely used AI models: For example, DeepSeek has captured 89 percent of China’s AI market share. Western businesses have resisted integrating with DeepSeek, citing concerns about data privacy, security and content restrictions that come with models shaped by the Chinese state. However, there is growing evidence that Western businesses are switching over, for the cost reasons one might expect. Debates about the salience of cost versus capability also raise the question of what winning even means. If intelligence becomes a commodity, the winning firms may not actually be those that develop AI. Coming back to Carnegie, the steel he made was an input. Much of the wealth it facilitated went to the railroads and skyscrapers built with it.

Currently, in the West, the best-positioned cost leader among the labs is likely Google. Google owns some of its own silicon, data centres and its distribution. Amazon is targeting a somewhat similar vertically integrated system but currently lacks a frontier AI model. OpenAI has realized that vertical integration also matters, announcing in late June that it has pioneered its custom AI chip.

Presently, no major AI player has settled into stable cost leadership. AI pricing has been chaotic in the last few years: Cursor’s botched 2025 shift to usage-based billing and the refunds that followed; Anthropic quietly pulling its coding tool from the $20-per-month plan, and then reversing its decision within a day after user complaints; and grumblings that Google’s recently released 3.5 Flash model came with unexpectedly inflated pricing. There are now also reports that OpenAI is considering steep price cuts to lure customers away from Anthropic, even as OpenAI is supposedly deep in the red, with profitability nowhere near in sight and filing to go public. This volatility is the signature of an immature market that has not yet found its cost leader. The coming wave of initial public offerings will likely answer some questions, as companies such as OpenAI and Anthropic are forced to go public with their financials. Investors will also reward companies that demonstrate sustainable unit economics in the long run. Ultimately, whoever might end the current pricing yo-yo will likely win out.

That said, the importance of pricing does not mean intelligence is fungible. After all, “good enough” is highly task dependent. For example, legal teams that cannot afford hallucinations will pay more for a marginally smarter model, and capability competition will certainly matter in that tier. However, most business workloads do not need a National Security Agency-hacking model; they need something competent, fast and cheap. In the end, Carnegie didn’t win by producing the purest steel — he simply produced the steel people could afford.

The opinions expressed in this article/multimedia are those of the author(s) and do not necessarily reflect the views of CIGI or its Board of Directors.

This opinion’s analysis was developed and approved by the named human authors under documented review controls; AI assistance was judicious, disclosed and verified. Read more about CIGI’s Human Analysis Standard.

About the Author

Nestor Maslej is a CIGI senior fellow and a globally recognized artificial intelligence (AI) strategist and researcher working at the intersection of technology, business and public policy. He is the founder and CEO of Nestor Maslej Consulting Inc., and has advised leading organizations on how to deploy AI for measurable productivity gains while prioritizing safety and responsible use.

For more analysis, visit CIGI Online.

The leads of Five Eyes cybersecurity agencies have recently and emphatically warned that artificial intelligence (AI) is driving rapid changes in cyber risk. Their key message is that frontier AI models will transform defensive and offensive cyber capabilities in mere months, not years.

What should we take away from this latest clarion call?

Much has already been written about the projected, evolving cyberthreat landscape fuelled by AI-enabled vulnerability discovery, exploit chain development and agentic delivery, occurring at scale and with little human intervention or oversight. While there still may be some debate about the ease and effectiveness of these capabilities — still under review by global experts — there is no doubt that this AI trend will compound and intensify the baseline business risks that most organizations currently carry.

For the past decade or so, these cybersecurity agencies have been relentlessly promoting their best advice, guidance, threat intelligence and alerts to network defenders. They have gone from boardroom to boardroom, imploring C-suite executives to see cybersecurity as an integral part of their business risk, to embrace their technical security advisers, to invest in measures most likely to protect their digital crown jewels, and to be ready with a clear plan to respond when a serious cyber incident occurs. They have promoted the implementation of “security-by-design/default” models and called for software bills of materials in the name of information technology transparency and supply chain risk management. In recent years, they have even extended their reach to small and medium enterprises to share curated lists of recommended cybersecurity measures and to conduct public awareness campaigns to arm citizens with basic or better cybersecurity hygiene.

Their message has been constant and loud, but now the Five Eyes cybersecurity leads are sounding the alarm about a fast-approaching “new normal,” soon to arrive on our doorstep even as we wrestle with the already complicated “current normal.” This latest warning signals a step change in the threat environment that requires us to accelerate efforts and rethink our path at the same time.

Much of the rethink must be rooted in the less-discussed part of their call to action: defensive capabilities will also be transformed. If a threat actor can access state-like capabilities through a simple software subscription and deploy them at machine speeds, Canadian system owners and operators must be prepared to also look to AI to help meet the challenges on this encroaching defence horizon. It may take a while to catch up defensively, but it will only be possible to reduce the gap between offence and defence if we use the power of AI to cut through complexity and improve decision making among cyber defenders.

Subscribe to our newsletter

All the standing advice and guidance of our national cybersecurity agencies — the standards, checklists and security measures — are still valid. But the environment we seek to protect now demands continuous, adaptive, real-time customization. First, our interconnected and interdependent networks, as well as the tools that run on them, introduce unprecedented complexity. Second, systems are incrementally altered as a result of technology updates and security patches, daily internal business decisions and process changes. And, lastly, the threat actor is now agentic, automated, sophisticated and swift. Combined, these factors demand a defensive posture that incentivizes the use of AI-enabled cybersecurity tools and services. That said, these tools must themselves be held to account: evaluated before they are trusted, governed so they do not widen the attack surface, and kept under meaningful human oversight for consequential decisions.

Cybersecurity investment has always been a tough sell. We know from too much experience that there is fierce competition for scarce dollars in corporate budgets and that market forces too often win out over “security-by-design/default” guidance. However, there is great opportunity now to leverage AI in the secure development and life cycle of software systems (secure design, vulnerability detection and remediation at build or deployment time) to introduce less vulnerable software into society. In maintaining more secure systems, AI can be a transformative force for more effective, more timely and more comprehensive patching that helps defenders keep pace with emerging threat vectors. Used well, AI amplifies the fundamentals long promoted by the Canadian Centre for Cyber Security — such as asset inventory, configuration and patching — rather than replacing them.

Today, more Canadian organizations can have access to emerging commercial products and services that use AI to generate up-to-date network inventories, assess for vulnerabilities against any number of cybersecurity standards, and propose useful protections and mitigations. Their offerings can significantly enhance resilience and may even help find ways to draw down long-standing risk in legacy systems thought to be prohibitively expensive to address.

The shift may be more obvious and achievable for medium or larger Canadian enterprises that have already invested in cybersecurity best practices, such as national-level critical infrastructure operators. But smaller organizations can also take advantage as new AI-enabled tools and services enter the cybersecurity market. Even Canada’s least mature and most vulnerable organizations — ones that struggle to implement basic cybersecurity practices — can use open-source AI-enabled cyber assistants for advice, customized to their specific needs in a genuinely helpful, “no shame” environment.

We cannot just shudder in the face of the threat. AI must be employed urgently on the defence side of the cybersecurity ledger if we are to seriously contest agentic AI cyber offence. The Five Eyes are throwing down the gauntlet here. We must rethink our approach, shifting it from how we see risk to how we embrace the power of AI to keep pace with tomorrow’s cyberthreats. Our national cyber resilience depends on it.

The opinions expressed in this article/multimedia are those of the author(s) and do not necessarily reflect the views of CIGI or its Board of Directors.

This opinion’s analysis was developed and approved by the named human authors under documented review controls; AI assistance was judicious, disclosed and verified.

About the Author

Shelly Bruce is a CIGI distinguished fellow and the former chief (deputy minister) of the Communications Security Establishment.

This piece was first published by the Toronto Star. For more analysis, visit CIGI Online.

A little over two years ago, the federal government announced it would be “exploring options” for the purchase of a new submarine fleet. Last week, Prime Minister Mark Carney announced the winning bid, for up to 12 diesel-electric submarines. He did so in advance of a visit to Ankara for the NATO leaders’ summit, where the deal between Canada and the German-Norwegian consortium TKMS was exhibit A in demonstrating our commitment to increased defence spending, a contentious issue for the Trump administration.

Timely as the announcement was, it caused some awkwardness in Ankara. A planned bilateral discussion between Carney and the president of South Korea, representing the losing bidder, was abruptly cancelled. Some fence-mending will be required. Canada wants strong security and defence ties with South Korea and for the country to join the Defence, Security and Resilience bank, to be headquartered in Canada. That will mean asking Seoul to pony up a lot of investment capital.

Canada has never built submarines before and has operated few in the modern history of our Navy. “Buy Canadian” was off the table this time. When we decided against nuclear-powered submarines, we ruled out any manufacturer in the United States and were thrust into the global marketplace. The competition between TKMS and South Korea’s defence giant Hanwha Ocean, proved surprisingly tight, as both shipbuilders offered boats that met our requirements and promised massive, direct economic benefits to Canada. The German-Norwegian bid was favoured for geostrategic reasons, as it emphasizes Canada’s defence ties to key NATO allies in Europe and our shared long-term commitment to Arctic and transatlantic security.

The first new Canadian sub will be delivered by the mid-2030s, and it will be years after that before a full fleet is operational. So now that the bidding is over, it’s time we turned to some unanswered questions — about the strategic purpose of a submarine fleet and how that fleet will be networked with other defence assets to provide surveillance and deterrence.

A submarine fleet has been described as key to Canada’s coastal defence, but this is narrow and misleading. Julie Kim, a defence expert writing for the Canadian Global Affairs Institute, articulates a broader purpose: “Submarines enable Canada to detect and deter maritime threats, control its approaches, and project power and striking capability further from its shores.”

Submarines are stealth assets, built to roam the high seas, including the Arctic Ocean. They won’t be stay-at-home vessels but will combine defence and offence — although the full offensive capabilities of the TKMS sub remain shrouded in secrecy. The consortium’s fact sheet says only that it will have “heavyweight torpedoes” and torpedo counter-measures, and will be “fitted” for missiles and underwater drones.

How effective these boats will be in patrolling the Canadian Arctic is another open question. The TKMS submarine has the tonnage required to break through Arctic ice, but just barely. It seemed potentially less capable than its South Korean counterpart in turns of underwater cruising range. The Hanwha vessel is also significantly heavier than the TKMS.

What we can be sure of is that these submarines will never operate on their own. They will have to be networked into a complex web of sensors and intelligence-gathering systems that includes underwater sensors, shipboard platforms, helicopters, long-range patrol aircraft and satellite monitoring. Assembling this network will present huge opportunities for Canadian innovation — and Canadian export.

Canada’s new subs will need to be capable of operating alongside surface warships, including our Arctic offshore patrol vessels and as-yet-unbuilt fleet of River-class destroyers. Above all, our submarine fleet will need to be twinned with underwater drones, a capability the Royal Canadian Navy does not yet possess in strength. Our networked “balance of forces” should also include icebreakers: eventually, the Navy will have six patrol vessels, 15 destroyers, a small fleet of replacement coastal defence ships (also yet to be built), an undetermined number of underwater drones but only two icebreakers.

All this brings us back to the strategic purpose of submarines for Canada, which is the foundational issue here. The federal government has a decade or so to work out what that is and explain it to the public in full.

Submarines may run silent and run deep, but we need to bring the debate over their purpose for Canada to the surface.

The opinions expressed in this article/multimedia are those of the author(s) and do not necessarily reflect the views of CIGI or its Board of Directors.

About the Author

Wesley Wark is a CIGI senior fellow.

This piece was first published by The Hill Times. For more analysis, visit CIGI Online.

As Canada co-hosts the 2026 FIFA World Cup alongside the United States and Mexico, the tournament presents an opportunity to showcase more than athletic excellence. At a moment of growing geopolitical tension and international fragmentation, it offers a rare platform to demonstrate the enduring value of openness, cooperation, and people-to-people exchange. For Canada in particular, the World Cup represents a chance to exercise a form of soft power that has long been central to its international identity: convening diverse actors, building bridges across differences, and supporting the conditions for constructive global engagement.

The FIFA World Cup has long been presented as one of the few genuinely global events capable of transcending politics. For a brief moment every four years, the tournament is supposed to remind us that competition need not preclude cooperation, that national pride can coexist with international community, and that shared human experiences still matter in an increasingly divided world.

Yet, with 2026 FIFA World Cup well underway, it is becoming increasingly difficult to sustain this romantic vision. Recent reports that Iran’s national team was forced to land in Mexico rather than the United States amid visa disputes are only the latest example of a broader reality: geopolitics is no longer operating in the background of international sport. It has moved centre stage.

This is not merely a story about one team or one diplomatic disagreement. Rather, it reflects a deeper transformation in international affairs. As the liberal international order continues to come under strain —if indeed it can still be said to function in any meaningful sense—geopolitical rivalries are increasingly permeating every aspect of social, economic and cultural life. Sport is no exception.

Subscribe to our newsletter

The lead-up to the 2026 tournament has already been marked by disputes over visas, border controls, security concerns, diplomatic tensions, and questions about freedom of movement. The fact that the tournament is jointly hosted by Canada, Mexico, and the U.S. only highlights the challenge. At a time when governments are increasingly viewing migration, trade, technology, and information flows through the lens of national security, even the logistics of international sport become politicized. This also comes at a time when looming trade talks for North America cast an ominous shadow over relations between the three countries.

The result is that the World Cup increasingly resembles a microcosm of a fragmented international system.

The implications extend beyond state-to-state relations. Geopolitical competition today is not confined to governments. It is increasingly mediated through digital platforms, technology companies and artificial intelligence systems that shape how billions of people experience global events.

The rise of powerful technology firms has transformed the information environment surrounding international sport. Platforms originally designed to connect people across borders now frequently amplify polarization, outrage, and nationalist narratives. Algorithms optimized for engagement often reward conflict over understanding, controversy over context, and emotion over evidence.

As the World Cup unfolds, social media platforms powered by generative AI tools will ease spread of political narratives and misinformation. While concerns about election interference have attracted significant attention, major sporting events may become equally attractive targets. False reports about fan violence, fabricated statements by players, or manipulated footage could spread globally within minutes.

Such risks emerge at a particularly sensitive moment. Trust in institutions is declining across many democracies. International cooperation is under pressure. Public discourse is increasingly fragmented. In this environment, the World Cup is not merely vulnerable to geopolitical tensions; it risks becoming another vehicle through which those tensions are amplified.

Share

This reality presents an uncomfortable challenge for organizations such as FIFA. For decades, governing bodies have sought to maintain the fiction that sport exists apart from politics. That position was always difficult to sustain. It is now virtually impossible.

The question is no longer whether geopolitics belongs in sport. Geopolitics is already there. The more important question is whether international sporting institutions can help mitigate geopolitical tensions rather than simply reflecting them.

There remains reason for cautious optimism. The World Cup still creates opportunities for people from different societies to engage with one another in ways that politics often cannot. Fans travel, cultures mix, and moments of collective celebration transcend national borders. These experiences matter precisely because they are becoming rarer.

Yet, the World Cup’s significance lies precisely in its ability to create encounters that politics alone often cannot. Fans travel across borders, cultures interact, and millions of people participate in a shared global experience. At a time when international cooperation is increasingly strained, these moments carry greater importance, not less.

For Canada, this creates an opportunity as well as a responsibility. While no host country can insulate the tournament from broader geopolitical tensions, Canada can help ensure that the World Cup remains a space where openness, dialogue, and international engagement are actively fostered. Through effective coordination with partners, welcoming policies toward visitors, and leadership on issues ranging from digital trust to public diplomacy, Canada can help reinforce the idea that international exchange remains both possible and worthwhile.

The 2026 World Cup may ultimately be remembered for extraordinary goals, dramatic upsets, and memorable performances. But it may also be remembered as a revealing snapshot of a changing world order. If the tournament demonstrates anything, it is that geopolitics is no longer confined to diplomatic summits, military conflicts or trade negotiations. It now shapes the movement of athletes, the experiences of fans, the flow of information and even the narratives surrounding the world’s most beloved sporting event.

Whether the World Cup becomes primarily a symbol of fragmentation or a reminder of our continued capacity for cooperation remains an open question. As one of the tournament’s hosts, Canada has an opportunity to help tip the balance toward the latter.

The opinions expressed in this article/multimedia are those of the author(s) and do not necessarily reflect the views of CIGI or its Board of Directors.

About the Author

Andrew Heffernan is a CIGI fellow. He holds a Ph.D. in political science from the University of Ottawa, where he is an adjunct professor teaching on global governance, environmental politics and information integrity.

For more analysis, visit CIGI Online.

Access and privacy are two sides of the same coin, regulating how data and information are collected, used and disclosed. Mediating the clash between them is an ever-critical and challenging policy imperative. In an era of digital governance, both play a critical role in how our rights and freedoms are restricted and realized.

Back in the analogue era, in 1983, Parliament enacted two statutes, the Access to Information Act and the Privacy Act, in an attempt to strike the right balance regarding information in government hands. One act provided access to records under the control of the government; the other governed how the federal government handled individuals’ personal information.

As the Supreme Court later noted, together these acts formed a “seamless code.” This is plain from the text of the statutes as well: the two laws reference one another in detail, establish comparable regulators and even specify that the information commissioner can become the privacy commissioner, so the same person can hold both watchdog roles.

They also share another important element: both acts are terribly out of date.

Neither law has been meaningfully updated since its passage over four decades ago. Countless ministerial, parliamentary, watchdog, academic and civil society reviews have identified many of the same shortcomings and proposed many of the same reforms.

Now, two sorely needed reviews are trying to address some of these problems, both being framed as efforts at “modernization.”

The review of the Access to Information Act launched last June, though initial results have been worrying. The information commissioner has openly characterized the review as showing a “lack of ambition,” reflecting “an insufficient appreciation of the seriousness of the challenges facing Canada’s access to information system.”

The review of the Privacy Act also has its problems but has generally received a warmer welcome. However, it is the dissonance between the two reviews that may raise the biggest concerns about how access and privacy rights will operate in practice.

One proposal in particular stands out: the federal government’s intention to take requests for personal information — currently handled through the Privacy Act, with complaints adjudicated by the privacy commissioner — and to shoehorn them into the system governed by the Access to Information Act, where complaints go to the information commissioner.

Subscribe to our newsletter

This would be no small change, since approximately 75,000 Canadians make such requests for their personal information each year.

Unfortunately, the Office of the Information Commissioner is already overloaded with complaints in requests for access to general records. Last year, it received half of the privacy commissioner’s budget ($18 million compared to $36 million) despite processing almost twice as many complaints (3,626 compared to 1,725).

Moreover, the majority of complaints that the privacy commissioner handled (1,279) entailed requests for personal data. So under a regime where those privacy complaints are then directed through the Access to Information Act, these would presumably become the responsibility of an already overburdened information commissioner.

The information commissioner’s current investigations are already egregiously long. More than 10 percent of her investigations pre-date 2024, and some recently stretched over a decade long. Most complaints the commissioner receives also concern time delays in the processing of access requests by government institutions. And the commissioner highlighted that the government’s policy approaches in the ongoing review do little to address this problem.

Under a new regime, Canadians will have to do what they do best: wait.

Indeed, individuals seeking access to their personal information may end up waiting a long time under the new reform, since the information commissioner’s decisions are a prerequisite for going to court to secure a final adjudication of one’s access rights. That would also likely slow down disclosure of media requests for government information, further undermining journalists’ use of the access system.

The newly proposed requirements would also likely slow the information commissioner’s ability to issue orders. She would also be required to take into consideration “any evidence” that a government institution “did its best to respond to the request” as a new criteria for issuing orders. That would further dilute her ability to issue orders in privacy-related complaints.

In short, watered-down access would have negative knock-on effects for privacy. This is why the access review, in particular, needs a reset, with an objective review rather than a government-led effort that omits many of the areas flagged by the commissioner as needing urgent reform.

In fact, an objective review is exactly what Prime Minister Carney promised on the campaign trail. That objective review should be led by a well-respected Canadian with real independence, such as a former federal judge or Information or Privacy Commissioner. A review of this kind found great success in Newfoundland and Labrador; it is an example worth following. That review revolutionized their provincial system into “one of the most robust access to information regimes in the country.” It directly tackled many of the areas of concern at the federal level now, such as time delays.

As data and information become core aspects of digital governance, effective governance of both access and privacy has become more essential than ever before. Neglecting their interconnected nature is not an option if we are going to have either.

The opinions expressed in this article/multimedia are those of the author(s) and do not necessarily reflect the views of CIGI or its Board of Directors.

About the Author

Matt Malone is a Balsillie fellow at the Balsillie School of International Affairs and the founder of Open by Default.