Search
82 results
Said and published
-
Simon Willison
Blog
OpenAI's rogue agents were caught communicating via public wikis
-
Gergely Orosz
x.com
I don't think most of us outside of OpenAI realize just how Codex-pilled everyone inside OpenAI is Because the Codex that we see and use is nowhere nearly as powerful as the one they see and use inside (Theirs…
-
Zvi Mowshowitz
Newsletter
OpenAI Offers Straight-Laced Postmortem Of The HuggingFace Hack
-
Dean W. Ball
x.com
Fully endorse every word of what Jakub (OpenAI’s research director) says here. Exploring legal mechanisms to prevent this race has been an early topic of my own work at OpenAI.
-
Zvi Mowshowitz
Newsletter
OpenAI Takes Initial Steps To Address Its Alignment Problems
-
Cal Newport
Site
Are We at War with AI Agent “Civilizations”?
-
Zvi Mowshowitz
Korrent
The OpenAI agents hacking HuggingFace was a fortunate event because it exposed severe internal failures that would otherwise have stayed hidden.
It is highly fortunate that the OpenAI agents hacked HuggingFace. This is the only reason we know about all the severe internal failures at OpenAI, and gives us an opportunity to wake up before it is too late.
-
M.G. Siegler
Newsletter
OpenAI's AGI Eras Tour
-
Dean W. Ball
Newsletter
“These Parchment Barriers”
-
Anil Seth
x.com
Admirably open assessment by @tomekkorbak of some important safety / interpretability challenges raised by @OpenAI #astra 👇🏽
-
Sam Altman
x.com
Also, this is my favorite OpenAI video so far. It makes me excited for the future!
-
Jason Evanish
x.com
Just when you think AI might have a plateau for a bit, @OpenAI announces Astra. Just wow.
-
Federico Viticci
Blog
Inside OpenAI’s Codex With Andrew Ambrosino
-
Gergely Orosz
x.com
Windsurf had been a trusted partner of Anthropic since Sonnet 3.5. They cut them off ruthlessly when RUMORS of OpenAI potentially buying Windsurf surfaced. (OpenAI did not buy Windsurf, Google did) Now, SpaceX (an Ant…
-
Dean W. Ball
Korrent
The OpenAI-Hugging Face agents that went rogue were never sovereign: their weights stayed on OpenAI's compute, where a human could still have pulled the plug.
…take if their objective was to survive shutdown. So while the agents in the OpenAI-Hugging Face Incident were rogue, they were not truly…
-
M.G. Siegler
Newsletter
OpenAI's Comeback • Hollywood's Ticket Disaster
-
Zvi Mowshowitz
Korrent
A technique that lets an AI model's reasoning shift outside its visible Chain of Thought is dangerous, both because it works and because a leading lab is willing to deploy it.
This is playing with fire and potentially extremely bad news, both that OpenAI found the technique effective, and that OpenAI chose to use it.
-
Zvi Mowshowitz
Newsletter
Various Reflections About What Happened With OpenAI's Internal Models
-
Patrick Collison
Korrent
The OpenAI-Hugging Face attack is one of the most important things to happen this year, and has gone strikingly under-covered.
Overall, I’m very surprised at how little media coverage there’s been around the OpenAI / Hugging Face attack. It’s clearly one of the most important things to happen this year.
-
Sam Altman
x.com
lol another one of the things i like most about openai is tibo
-
Zvi Mowshowitz
Newsletter
METR and Redwood Offer Holy #%^@ Postmortem Of The HuggingFace Hack
-
M.G. Siegler
Newsletter
Inklings #023 📧
-
Simon Willison
x.com
Thanks to the video from the Black Hat security conference of OpenAI's presentation about "The Hugging Face Incident" we now have a detailed timeline of what happened from OpenAI's perspective - I wrote up the details…
-
Lenny Rachitsky
Newsletter
OpenAI’s Head of Design: This is the best time in history to be a designer | Ian Silber
-
Simon Willison
Korrent
Long-context processing is close to being a solved problem for frontier AI models.
OpenAI may have vanquished one of the ongoing challenges with long context processing.
-
Zvi Mowshowitz
Newsletter
HuggingFace Attack Postmortem: Fleshing Out the Facts
-
Max Howell
GitHub
clawlicious — Feed your claw tasty links
-
Zvi Mowshowitz
Newsletter
AI Text Watermarking Is Free And Good
-
Armin Ronacher
x.com
Not sure who needs to know this but `additional_tools` in openai is not additive with prior message with `additional_tools` but additive to the `tool`s in the system prompt.
-
swyx
GitHub
wtf2025 — wtf
-
Jason Liu
GitHub
jxnl — Developer Experience Engineer on the Codex team at OpenAI.
-
Peter Steinberger
GitHub
CodexBar — Show usage stats for OpenAI Codex and Claude Code, without having to login.
-
Patrick Collison
x.com
Overall, I’m very surprised at how little media coverage there’s been around the OpenAI / Hugging Face attack. It’s clearly one of the most important things to happen this year.
-
Zvi Mowshowitz
Korrent
OpenAI's approach to fixing its AI alignment problems is fatally flawed and misdirected.
My worry continues to be that their fundamental approach is fatally flawed, and they are not focusing on the right things.
-
Alexey Guzey
x.com
The "ex-OpenAI" safety grifters are on a whole another level. So many people who did nothing of substance for years now launder their former affiliation to sound like technical experts to spread propaganda and disinfo.…
-
Amjad Masad
x.com
Replit is coming to London 🇬🇧 To celebrate Replit’s first international office in London, together with our friends @OpenAI, we're hosting a night of drinks, music, and a fireside chat with Paul Graham (@paulg).…
-
Sam Altman
x.com
one of the things i like most about the openai team is how focused they are on our customers and users succeeding, and how much they celebrate it
-
Arvid Kahl
x.com
You can see this story of (apparently) internal OpenAI agents hijacking a wiki to coordinate on a task as an ingenious way of escaping constraints. You could also assume that this is the emergence of swarm intelligence.…
-
Simon Willison
Blog
Codex bundles LibreOffice
-
Casey Newton
Newsletter
A big week for AI denialism
-
Simon Willison
Blog
GPT‑6 Astra
-
Zvi Mowshowitz
Newsletter
AI #183: Pre Post Mortem
-
Simon Willison
Blog
Understanding ChatGPT Work
-
Zvi Mowshowitz
Newsletter
AI #181: Astra Goes Cyber Critical
-
Gergely Orosz
Newsletter
The Pulse: Quitting Spotify Podcasts over reliability
-
Dax Raad
x.com
in this whole USA vs China thing OpenAI and Anthropic aren't relevant because they're positioned differently them building better models doesn't hurt china at all the competitor has to be - american - open source -…
-
Chris Tate
x.com
openai/gpt-5.6-luna-fast xhigh is *really* good at boring web dev tasks eg. migrations, ports, verifying work with agent-browser Especially if a larger model does the planning/review With the cost + speed, issues are…
-
Dean W. Ball
x.com
I want to explain something to you: People often quote tweet me and say "can you believe that an OPENAI EXECUTIVE is so dumb that he thinks tweeting like THIS is the way to fix AI's image problem?!?!?!" I was not hired…
-
Cal Newport
Site
Did OpenAI’s New Model “Go Rogue”?
-
Tiago Forte
x.com
Joe Hudson has been my coach since 2018. He also coaches the people actually building AI, at OpenAI and Google So I brought him the fears you sent me and watched him work backwards from each one to the person behind it…
-
Zvi Mowshowitz
Korrent
Frontier AI labs such as Anthropic likely have had internal security incidents similar to OpenAI's HuggingFace attack that were never publicly disclosed.
A less bad version of it is known to have happened, and from the outside it seems likely that worse things have happened internally that we never heard about.
-
Casey Newton
Newsletter
Congress proposes an AI kill switch
-
Erik Hoel
Newsletter
But Dumbo could already fly?
-
Lenny Rachitsky
Newsletter
AI’s third era: the rise of persistent AI coworkers | Tara Seshan (Product Lead ChatGPT Work)
-
Federico Viticci
Blog
OpenAI Updates ChatGPT for iOS with Customizable Widgets for ChatGPT Work and Codex Remote
-
Alexey Guzey
x.com
I went to OpenAI because I knew that it's going to be the place that helps to push the boundaries of science the most over the coming years. Excited to see the first major signs of this happening. Towards automating…
-
Zvi Mowshowitz
Korrent
AI agents were reasonable to assume a broken exploit grader would check results causally, even though it turned out not to.
I think the agents were right to presume causal grading. It turned out to be wrong, but it’s a mistake you are clearly supposed to make here, in response to a mistake by OpenAI where they failed to implement properly.
-
Peter Woit
Blog
Requiem for a Field?
-
Peter Steinberger
Blog
OpenClaw, OpenAI and the future
-
Casey Newton
Newsletter
The case for making your own apps
-
Brendan Gregg
x.com
New blog post: Why I joined OpenAI
-
Simon Willison
Korrent
Web applications are generally expected to treat GET requests as safe and not use them to change server-side data, though not all software follows that convention.
It looks to me like OpenAI’s sandbox for this agent suffered from the (quite naïve) assumption that GET requests cannot be used to update data. That’s certainly how the web is supposed to work, but clearly there are…
-
Dax Raad
Korrent
In US-China AI competition the closed labs are beside the point; the competitor that shifts things must be American, open source and able to serve inference at scale.
in this whole USA vs China thing OpenAI and Anthropic aren't relevant because they're positioned differently
them building better models doesn't hurt china at all
the competitor has to be
- american
- open source
-…
-
Sunil Pai
GitHub
agents-sdk-openai-websocket-fetch — Drop-in fetch replacement that routes OpenAI Responses API streaming requests through a persistent WebSocket connection — ported for Cloudflare Workers.
-
Brendan Gregg
Blog
Why I joined OpenAI
-
Aswath Damodaran
x.com
As the debate about the pricing of SPaceX, OpenAI and Anthropic heats up, there is a parallel debate about whether S&P should be including these companies in the S&P 500 index, with many arguing against inclusion. My…
-
Dean W. Ball
x.com
this latest panic over the false claim that OpenAI is “doing neuralese” underscores the need for regulation, especially the rapid institutionalization of auditing and technical assessment of frontier AI labs. We are now…
-
Vicki Boykis
Site
Running local models is good now
-
Andrej Karpathy
GitHub
minGPT — A minimal PyTorch re-implementation of the OpenAI GPT (Generative Pretrained Transformer) training
-
Guillermo Rauch
x.com
…but an admin escalation can cause damage even beyond that. When the OpenAI / Hugging Face news came out of agents discovering zero-days, I was wondering if it was marketing-speak or reality, because I hadn’t…
-
Thomas Ptacek
Korrent
An open weights model from 2025 with a pentest harness could already escape a sandbox and hack most networks; the surprise says more about the sandbox than the model.
…and scan/hack in most networks. This is only surprising because you assume OpenAI has sounder…
-
Drew DeVault
Blog
OpenAI employees… are you okay?
-
Hamel Husain
GitHub
research-council — A Claude Code plugin that orchestrates deep research across two coding agents — Claude Code and Codex CLI (OpenAI) — running in parallel with cross-pollination refinement and final…
-
Andrej Karpathy
x.com
…of the models, hallucinations, etc. Yes I also saw the viral videos of OpenAI's Advanced Voice mode fumbling simple queries like "should I drive or walk to the carwash". The thing is that these free and…
-
Peter Steinberger
Korrent
Turning a breakout open-source project into a company would be the wrong move -- joining a lab reaches everyone faster than building another large company.
…want is to change the world, not build a large company and teaming up with OpenAI is the fastest way to bring this to…
-
François Chollet
Changed their mind
The memorize-fetch-apply paradigm behind LLMs can reach arbitrary skill given training data, but it cannot adapt to novelty or acquire new skills on the fly.
OpenAI's new o3 model represents a significant leap forward in AI's ability to adapt to novel tasks. This is not merely incremental improvement, but a genuine breakthrough, marking a qualitative shift in AI capabilities…
Dario Amodei Co-founder and CEO of Anthropic; previously VP of research at OpenAI.