Where the AI market is talking

A misalignment of AI in mathematics

Unless you’re very weak-willed, I wouldn’t worry about independence. LLMs make mistakes all the time, they are nothing like an all-knowing oracle nor are they going to replace humans despite the absurd fantasies of LLM fans and those with a vested interest. Plagiarism is an interesting point, though honestly I think it would be fairly easy to work out who had published similar research if the LLM gives you an idea - personally I see this as the weakest argument against using them, as long as you are strict about attribution - all work like this depends heavily on the research of others - the LLM is just a tool to aid that research IMO. Opsec is a fair point, and it might be worth avoiding the completely amoral OpenAI at this point for that reason. There are open models though.

OpenAI's Navier-Stokes claim splits the math community

OpenAI 20x Pro subscription plans temporarily unavabilable

A misalignment of AI in mathematics

The anthropic principle applies here: anyone warning about an existential risk will by necessity never have precedent to point to. I am not assuming humans don't remain in control, but rather, I am not axiomatically assuming there's any inherent reason we will.

A misalignment of AI in mathematics

>> I’m no big AI cheerleader, but why don’t you try these models to see what they offer - they might suggest some things you haven’t thought of, or save you some time in your research. One reason not to try is to create an air gap between the OP's ideas and the data that future models can train on. We saw that the mathematicians who trusted OpenAI and Anthropic with their preliminary work found the rug pulled out of their feet by those same companies. Another reason is to avoid inadvertently plagiarising the work of other mathematicians. Any mathematical insight that comes out of an LLM is the result of training on the entire bibliography of mathematical research, but those insights are spat out of the models without attribution. If you use AI in your matematical research you are on…

A misalignment of AI in mathematics

>> I’m no big AI cheerleader, but why don’t you try these models to see what they offer - they might suggest some things you haven’t thought of, or save you some time in your research. One reason not to try is to create an air gap between the OP's ideas and the data that future models can train on. We saw that the mathematicians who trusted OpenAI and Anthropic with their preliminary work found the rug pulled out of their feet by those same companies. Another reason is to avoid inadvertently plagiarising the work of other mathematicians. Any mathematical insight that comes out of an LLM is the result of training on the entire bibliography of mathematical research, but those insights are spat out of the models without attribution. If you use AI in your matematical research you are on…

Navier-Stokes Announcement

1) OpenAI and Buckmaster did not solve the same problem. Per https://x.com/IlinVasily29521/status/2097554700321329393 , here is a breakdown of who solved what. Tristan + Levent: 3D incompressible Euler with forcing OpenAI: 3D incompressible Euler without forcing OpenAI: Navier-Stokes with forcing No one: Navier-Stokes without forcing Euler equations = Navier-Stokes without viscosity. Forcing means external force. Absence of viscosity and presence of external force make blowup easier to construct. Tristan+Levent ticked the weakest case, OpenAI ticked the two next weakest, then the final case is unsolved. Only the last two are eligible for the Millennium Prize. The Navier-Stokes general case remains unsolved. Navier-Stokes has an extra viscosity term compared to Eul…

'Immature playground boasting': Mathematicians uneasy at OpenAI's latest scalp

Navier-Stokes Announcement

It's not a large prize though. OpenAI spent many multiples of the prize money in just a few days to get there and even if one solves a problem in the traditional way, that person is most likely already an accomplished professor at a reputable university where a million dollars doesn't mean as much as the eternal fame that comes with it.

Navier-Stokes Announcement

Relevant, but it's already rumored that OpenAI and Anthropic have made very significant progress on two more Millennium Prize problems. They're in a race to solve the next problem. OpenAI needs to solve another to shut down the (baseless) plagiarism allegations. Anthropic wants blood because OpenAI sniped the last one from one of Anthropic's researchers. It's a matter of pride for both companies. More results will come out soon.

Navier-Stokes Announcement

Relevant, but it's already rumored that OpenAI and Anthropic have made very significant progress on two more Millennium Prize problems. They're in a race to solve the next problem. OpenAI needs to solve another to shut down the (baseless) plagiarism allegations. Anthropic wants blood because OpenAI sniped the last one from one of Anthropic's researchers. It's a matter of pride for both companies. More results will come out soon.

Bernie's AI bill proposes to sentence AI developers to 20 years in prison

I didn't say the challenges would be the same.. You just indicated they were worthwhile for anyone who believes the anthropic principle is why they haven't seen a world with no nuclear treaties decades later.

Detecting and countering misuse of AI: September 2026

There is no end to the stupid nowadays. I remember OpenAI tweeting to about heads of state asking ChatGPT on policy decisions. There were stories about young hackers using Discord to coordinate attacks. Multiple military installations and an aircraft carrier was identified by smart fitness watches. At least on Russian general was killed by Ukrainian assassins tracking him via his smartwatch.

So you want to use OpenRouter?

If you require consistency for research, I wouldn't recommend relying on hosted inference without validating the provider/inference stack very carefully. For my contribution to OpenAI's gpt-oss red-teaming competition (published as arXiv:2510.01259), I initially used OpenRouter with DeepInfra and Together AI. The results were much too noisy to draw reliable conclusions from, and markedly different behavior through AWS Bedrock was the last straw. I ended up renting GPUs through vast.ai and running the model myself with vLLM.

So you want to use OpenRouter?

If you require consistency for research, I wouldn't recommend relying on hosted inference without validating the provider/inference stack very carefully. For my contribution to OpenAI's gpt-oss red-teaming competition (published as arXiv:2510.01259), I initially used OpenRouter with DeepInfra and Together AI. The results were much too noisy to draw reliable conclusions from, and markedly different behavior through AWS Bedrock was the last straw. I ended up renting GPUs through vast.ai and running the model myself with vLLM.

GoatCode – open-source terminal AI agent with provider failover

GoatCode is a single 85 MB binary that runs in your terminal. It brings 180+ LLM providers (or your existing Claude/ChatGPT/Gemini/Copilot subscriptions via OAuth), and when your quota dies mid-session it automatically retries then walks your fallback chain live — switches provider mid-turn, says so, and finishes the answer. No crash, no lost context. What’s new in v2.1.10: - Prompt caching on Anthropic (up to 90% cheaper long sessions; savings shown in /cost) - Inline unified diffs after every write/edit - /rewind — jump transcript + files back to any turn - Parallel subagents (up to 3 task calls fan out concurrently) - /search + goat search — full-text across every saved session - @file Tab completion, /model sonnet aliases - True context meter (r…

AI researchers debate how close we are to recursive self-improvement

I see, we're talking about different things. My thought experiment was along the lines of "Let's say I'm Anthropic and I want to significantly improve my frontier model's performance on, say, theoretical physics research. How do I build a fully autonomous process capable of constructing an eval that's somewhat outside the current capability in some useful direction (decided by the autonomous process itself)?" Would love to hear folks' ideas. :)

Anthropic researcher resigning citing superintelligence ' – is a marketing hoax [video]

Navier-Stokes Announcement

Completely misleading. This is all you need to read and understand for Anthropic's FLT formalization: import Mathlib import Theorems.Thm_fermat_last_theorem /-- Solution side: the same statement, binder for binder, proved by this tree's `fermat_last_theorem`. -/ theorem FLT_for_comparator (n : ℕ) (hn : 3 ≤ n) (a b c : ℕ) (ha : 0 < a) (hb : 0 < b) (hc : 0 < c) : a ^ n + b ^ n ≠ c ^ n := fermat_last_theorem n hn a b c ha hb hc /-- Mathlib's named proposition, by the one-line bridge from the elementary statement (the bridge is restated inline so that this file depends only on `Theorems.Thm_fermat_last_theorem`). -/ theorem FLT_mathlib_for_comparator : FermatLastTheorem := fun n hn a b c ha hb hc => fermat_last_theorem n hn a b c (Nat.pos_of_ne_zero ha) (Nat.po…

OpenAI agents carried out an undisclosed attack on RubyGems

Nah, if Anthropic can do it, so can OpenAI: > After finding this incident, we broadened our search to roughly 481 million transcripts—an intentionally wide net, consisting of all transcripts from our Frontier Red Team, many non-cyber evaluations, reinforcement learning (RL) environments, subagent logs, and more. We performed a first-stage scan of this group of transcripts for signs of internet access, such as public IP addresses and web addresses, and a second-stage scan using Claude to review the 9.2 million transcripts the first stage flagged for escalation. This scan re-identified the four incidents and found no other cases of similar or worse severity. https://www.anthropic.com/research/alignment-assessment-cybe...