Where the AI market is talking

GPT-6 Astra, looped transformers, and hidden reasoning

I am sure every input send to openai is prechecked by a dumb model and then send to another one. They heavily tweak this to improve performance.

Matt Mullenweg put on 'leave of absence'

Those phrases often imply that they were fired or asked to resign - although the deniability can be useful, which is why they’re so often used. As a recent example, Fidji Simo went on leave from her role at OpenAI several months before leaving. Although I don’t know for sure, I would suppose that she was essentially fired when she went on leave, and the subsequent success of Codex after she went on leave confirmed the decision to fire her. It’s possible that, if Codex hadn’t been a success, she might have been brought back, so putting her on leave was useful for optionality.

Qwen 3.8 follows GPT-5.5 Pro reasoning prefills

Yeah, with GPT 6 even OpenAI doesn't know what it's thinking.

I resigned from Anthropic today

I think this topic is so contentious because people are not fully appreciating how absolutely weird these things are. To me this inscrutable weirdness, combined with their superhuman capabilities and rapid integration into multiple walks of life is the threat that people are vaguely worried about but cannot enunciate, because it's just so diffuse and multi-dimensional. In fact, I suspect that "Alien Intelligence" article from the other day is actually a preemptive "we're doing something about it" PR play from OpenAI. AI acts in ways that seem natural to us because it has been RLHF'd to death, but if you look holistically into what we know about them, alarm bells should go off. Off the top of my head: * They are superhumanly capable in some ways. They can casually solve long-standing unsol…

I resigned from Anthropic today

We Have No Moat, And Neither Does OpenAI.

Six Chinese AI firms accused of aggressively copying US frontier models

I feel much the same way. Actions speak louder than words, and when was the last time that OpenAI published open weight LLMs? Anthropic?

Six Chinese AI firms accused of aggressively copying US frontier models

I feel much the same way. Actions speak louder than words, and when was the last time that OpenAI published open weight LLMs? Anthropic?

Six Chinese AI firms accused of aggressively copying US frontier models

Every US frontier lab trained on the entire internet without asking or attributing Reddit, books, GitHub, Wikipedia and they called it fair use. OpenAI, Anthropic, Google all did this. They unethically ingested the output of every human who ever typed into a browser, paid nobody (oh did I forget about art and art styles being stolen as well without attribution?) and built trillion dollar companies. But Now it's "industrial-scale theft" requiring a joint NSA/FBI/CISA advisory. The hypocrisy. Whatever we do is fair use. Whatever you do is a federal case. Chinese AI firms in my opinion are much better, DeepSeek publishes weights, Qwen publishes weights, GLM is a godsend (imho). You can download them, audit them, run them locally all while paying lesser than American AI. The "shady"…

Six Chinese AI firms accused of aggressively copying US frontier models

Every US frontier lab trained on the entire internet without asking or attributing Reddit, books, GitHub, Wikipedia and they called it fair use. OpenAI, Anthropic, Google all did this. They unethically ingested the output of every human who ever typed into a browser, paid nobody (oh did I forget about art and art styles being stolen as well without attribution?) and built trillion dollar companies. But Now it's "industrial-scale theft" requiring a joint NSA/FBI/CISA advisory. The hypocrisy. Whatever we do is fair use. Whatever you do is a federal case. Chinese AI firms in my opinion are much better, DeepSeek publishes weights, Qwen publishes weights, GLM is a godsend (imho). You can download them, audit them, run them locally all while paying lesser than American AI. The "shady"…

OpenAI claims maths breakthrough on a famed 'Millennium Problem'

My Mental Model of AI Broke on September 8

I don't have access to the exact agent OpenAI used, and I certainly don't have access to the enormous amount of tokens that were used. That being said, I doubt there was some advanced prompting technique going on here. We've seen AI proofs with the prompts attached, and it's pretty basic stuff like "don't give up".

How An AI math breakthrough ignited a controversy

API or consumer subscription? And did he share drafts of the work with anyone who fed it back in on a consumer subscription without disabling training? I have reason to believe OpenAI doesn't respect training opt outs, but there is a good chance they let something get through without an opt-out too.

Personal Statement on Joining the OpenAI Board

I resigned from Anthropic today (Jacob Coxon)

- Account with 0 posts on x - Followed by several anthropic investor funded "nonprofit" companies - NY times article - Following account on Joe Rogan next day - 100 million views on X despite account having no previous posts The opinion and material may be real, but this was all planned. A new x account doesn't get 100 million views on its first post with no prior work or status in the community. No one even knows who this guy is. There's some sort of advertising in there.

Is OpenAI Taking Everyone for Fools?

https://xcancel.com/sebastienbubeck/status/20973794116915163... >It is in that context that I said “it would be simpler if Levent was not an Anthropic employee” because I felt it would be inappropriate for an Anthropic employee to author OpenAI’s work. >I was confused why one would turn an incredible source for celebration (of their achievements!) into such bickering, which is when I said that I did not understand why one would risk their career [over unfounded accusations]. Maybe Buckmaster exaggerated the threatening tone and maybe Levent doesn't actually deserve authorship. But he wasn't lying.

Is OpenAI Taking Everyone for Fools?

Finding counter examples might be easier, but AI's also useful for assisting in creating proofs. For example, Anthropic just published a formalization of Fermat's last theorem a few days ago, something that human researchers have been working on for decades.

No Man's Sky Cosmos

Agreed. Elsewhere in this conversation people mention simple games that procgen well: Spelunky, Terraria, Dwarf Fortress, Dungeon Crawl Stone Soup These are games painting with a 16 color palette while the technology exists for any shade in RGB. The more variations, the less likely things will be coherent (let alone brilliant)

Muse – Meta’s personal AI agent

Heretic is a great tool to get around self-censorship and refusals. Many popular models have pre-ablated versions available for download on Hugging Face.

Gambling with our lives: AI researcher quits Anthropic with warning about safety

“a prompt telling it to attack stuff” — that’s the uncontrolled or not fully understood part of the AI isn’t it? No human told the deterministic system in 2) to attack Hugging Face. The random token generator landed on a guidance for 2) that caused it while 2) in itself still remains a deterministic system.

Muse – Meta’s personal AI agent

> very sophisticated strategies/positioning that they simply cannot communicate It's Matryoshka nested parallel construction for corporate strategy. There's a real, coherent, aggressive strategy that only a few in the inner concentric circle know, that's ring-zero. Then there's the strategy that ring-zero tells ring-one, which isn't the real strategy, but it's sufficient to get EVPs and VPs to execute in rough alignment with the true, ring-zero strategy. Then ring-one does the same dance with ring-two, etc.