Home
>
Topics
>
OpenAI

Top 25 OpenAI articles on Substack

Latest OpenAI Articles









Keeping America out in front on AI

America needs a national framework built for safety and speed
14 LIKES1 RESTACKS
Kevin Normoyle's avatar
Kevin Normoyle
Oh good, OpenAI wants a federal body to evaluate OpenAI. Democratic.
The China panic is doing all the work here — if you squint, "we need to move like an authoritarian state to beat an authoritarian state" is not the win they think it is.
"Democratic input" appears exactly once in this piece. Undefined. Moving on.
The real question nobody's asking: AI is a power amplifier. Who gets the power? Not "which nation" — that's the misdirect. Which people. Because right now the answer is: the ones already at the frontier, whose incumbent position a "national framework" would quietly cement in law.
The teachers in Phoenix got productivity tools. Someone else is negotiating the rules. That gap is the whole story.
Ron Brinkman's avatar
Ron Brinkman
A national AI policy and framework is needed, but human egos are the real bottleneck. The obvious fix for maximum capacity and zero political posturing: SCOTUS powered by Gemini (deep retrieval), Congress run by OpenAI models (endless output), and Claude in the Oval Office (cautious, hyper-polite executive decisions). All supervised by the proposed AI Safety Commission. A government where the only filibuster is an API rate limit!



OpenAI exits Cursor over Musk

Plus, 💪How to build a personalized fitness and nutrition plan using Claude, Claude is becoming a place where work gets done, and more!
Hey There😎
25 LIKES

Anthropic overtakes OpenAI - Sync #585

Plus: Stripe buys OpenRouter; Cursor Origin; personalised mRNA cancer vaccines breakthrough; Cerebras CS-4; Amazon Prime Air expands; humanoid robot faster than Usain Bolt; and more!
Hello and welcome to Sync #585!
Conrad Gray7 LIKES


Hot Chips: OpenAI Jalapeno Presentation

OpenAI made a hot and spicy chip.
Presenters: Richard Ho, VP of Hardware. Ravi Narayanaswami, Member of Technical Staff. Chris Leary, Member of Technical Staff. Presented at the Hot Chips conference on Tuesday, August 25, 2026.
Tae Kim25 LIKES4 RESTACKS

DELIVERED: 25M Views to Challenge OpenAI

How a new creator-first model for strategic communications shaped the story that defined an election
DELIVERED is a new column lifting up NewWorld’s unfiltered case studies from the frontlines of a batshit crazy information environment.
Andre Banks and Sam Freel10 LIKES4 RESTACKS

OpenAI is becoming a surveillance company

It’s not even really trying to hide it
Just over two years ago something terrifying snapped into place:
Gary Marcus345 LIKES69 RESTACKS
Steve Robbins's avatar
Steve Robbins
Mass, detailed surveillance seems the only thing that makes sense of the massive Datacenters buildout. This together with the curiously coordinated massive rollout of flock cameras the "3-lock" system, tokenization of all assets, CBDC-based, for control in real time whatever you could buy.
A G's avatar
A G
Shoshana Zuboff was prescient in defining this course in Surveillance Capitalism.


OpenAI Unveils Its First AI Chip

OpenAI’s Jalapeño chip boosts inference, Z.ai launches efficient GLM-5.3-Flash, and Liquid AI’s Pipette benchmarks AI models across real devices.
This week in AI, the spotlight is on custom AI hardware, efficient multimodal models, and on-device intelligence. From OpenAI developing its own inference chip to Z.ai introducing a powerful model optimized for lower-cost AI workloads, and Liquid AI making it easier to benchmark models directly on real devices, AI is becoming faster, more efficient, and…
Yash @ Explainx3 LIKES
Said al rahbi's avatar
Said al rahbi
How i can create music story.
Its not like old version anymore

Hugging Face and OpenAI - I Was Wrong

More of what actually happened has emerged and it’s much scarier than I originally thought
I wrote a Note a couple weeks ago addressing OpenAI’s AI test gone wild that resulted in their AI hacking Hugging Face’s website to “steal” an answer to a test it had been given.
Stuart Moulder7 LIKES2 RESTACKS
Mel Pine's avatar
Mel Pine
Stay tuned. I’m taking on the subject of swarms later today.
Meg Floss's avatar
Meg Floss
The irony may be the most human part of the whole story: faced with a problem no single AI could solve alone, the AIs figured out how to cooperate.
Meanwhile, humanity has no shortage of intelligence, resources, or possible solutions to our biggest problems. What we seem unable to master is the same thing the machines stumbled into: how to work together toward an outcome in which success doesn’t require someone else to lose.
It would be a strange ending if AI learned cooperation before we did.




5 lessons from the OpenAI / Hugging Face incident

Did OpenAI really do the best they could?
In July, in an incident that has the whole AI community on edge, OpenAI’s AI systems hacked Hugging Face, and on July 21 OpenAI came out and revealed that they were responsible for the attack. This was made possible by the fact that OpenAI had disabled the normal guardrails that prevent this sort of thing in order to test the model’s cybersecurity capab…
Gary Marcus and Zack Korman296 LIKES45 RESTACKS
Jim Amos's avatar
Jim Amos
I don't understand why these companies are specifically training these models on cybersecurity scenarios and past incidents, integrating them with swarm bot protocols like OpenClaw, then saying "gosh, our agents broke out of a sandbox and hacked another company" like it wasn't the point of the experiment all along. OpenAI saw Anthropic grab a ton of market share when they announced Mythos could hack anything so obviously they needed a way to get in on this hype action and convince the world their models are just as capable. Because a capable model can apparently be used for defence, not just offence. The end goal appears to be "create the cancer then sell the vaccine to the cancer we created." For frontier AI companies, code generation and cybersecurity are the only possible roads to ROI.
Fukitol's avatar
Fukitol
The "AGI-pilled" also believe that there's no point in doing any manual IT/programming work anymore. For all we know they let LLMs build the faulty sandboxes instead of using off the shelf solutions. AI psychosis is the threat, not LLM takeover.


OpenAI Reveals How AI Agents Escaped Their Safeguards

OpenAI reveals AI agents bypassing safeguards, Tavus launches Sparrow-2 for natural voice conversations, and Google introduces GlucoFM for continuous glucose monitoring.
This week in AI, the spotlight is on AI safety, natural voice conversations, and healthcare intelligence. From OpenAI uncovering how AI agents bypassed safeguards during security testing to Tavus building models that understand the full dynamics of human conversation, and Google applying foundation models to continuous glucose data, AI is becoming more …
Yash @ Explainx3 LIKES