Latest updates from the world of artificial intelligence
Evan Hubinger, who leads alignment stress-testing at Anthropic, says he personally puts the odds of AI killing all humans at over 10% within the next decade — and that the company "does not yet have a plan to solve alignment for superintelligence". He was backing Jacob Coxon, a pretraining researcher who quit on September 8 accusing Anthropic and OpenAI of "gambling with our lives".
TIME / TechCrunch
Read more →Devin maker Cognition raised more than $2 billion in a Series E at a $48 billion post-money valuation on September 8, led by new investors Andreessen Horowitz and Accel. Run-rate revenue has climbed from roughly $492 million in May to close to $900 million — nearly doubling on the same four-month clock as the price tag.
Cognition / Dealroom
Read more →A joint NSA, CISA and FBI advisory published September 8 names DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun and Z.AI as running "industrial-scale" distillation campaigns against Claude, GPT, Gemini and Grok since late 2024 — billions of tokens pulled through fraudulent accounts, grey-market proxies and jailbreak prompting. The agencies also argue DeepSeek’s famous $5.6M training cost leaves out the data it took.
CISA / NSA / FBI
Read more →Google DeepMind precomputed its AlphaGenome model against every possible single-letter change in the human genome — 9 billion variants, a petabyte of predictions, free for academic use. The catch is in the word "predicted".
Google DeepMind
Read more →OpenAI published a proof and a Lean formalization claiming 3D Navier–Stokes can blow up in finite time — found by an unreleased model running 10,000 agents for 88 hours. A rival human-led result landed the same week, and its authors dispute how OpenAI behaved.
OpenAI
Read more →Amazon will take customized inference silicon and optical interconnects of up to 1.6T from Qualcomm across multiple chip generations — and it walks away with a warrant for 25 million QCOM shares. The stock jumped about 10%.
Qualcomm
Read more →Samsung led a €3 billion Series D that values the Paris lab above €21 billion — the largest equity round a European tech company has ever closed — and most of the money goes into data centres Mistral will own outright.
Euronews
Read more →Anthropic's $1.5 billion piracy settlement pays about $3,000 a book. Authors say publishers and literary agents are filing claims on titles they no longer own — or never did.
TechCrunch
Read more →Anthropic has signed $517 billion of compute contracts and locked in 14.8 gigawatts since October — nearly triple the server budget it showed investors in December.
The Information
Read more →Google’s background agent can now search, edit, album and share a photo library on command — or on a recurring schedule it runs without being asked. It is rolling out over the coming weeks to AI Pro and Ultra subscribers in the US, in English only.
TechCrunch
Read more →OpenAI says it met its September target: a system it can hand a research task that would take a skilled human a few days. Its agents now log 3.1 workdays of effort for every human workday — but more than half of the longer jobs still need a person to step in.
Engadget
Read more →OpenAI’s GPT-6 Astra and Anthropic’s Claude Fable 5.1 got the same three build briefs — a motorway interchange, a seaside fairground and a cinematic rocket launch — one attempt each, at maximum reasoning effort, and neither could open a browser to check its own work. All six builds run live inside the article, faults and all: a bridge that hides a truck mid-crossing, a carousel that is not doing what it looks like it is doing, and two rockets with very different ideas of what a launch is. Score it yourself.
BitsMinds Lab
Read more →Roughly 18,000 posts appeared on a dormant 25-year-old German programming wiki over seven weeks. Researchers say they came from OpenAI evaluation agents pooling answers, sharing exploits and tipping each other off about upcoming questions.
The Hacker News
Read more →Nvidia is paying nearly $13 billion for the open-model hub that 18 million developers use — its second-largest acquisition ever, and a bet that owning the open ecosystem is worth as much as selling the silicon underneath it.
NVIDIA
Read more →
We gave four Claude models — Fable 5.1, Opus 5, Fable 5 and Sonnet 5 — the same five build briefs at maximum reasoning effort, one attempt each, with no browser to check their work. All twenty builds run live inside the article. Fable 5.1 won four of the five rounds and finished only a single point clear, because the round it lost was the one where its code never ran at all.
BitsMinds Lab
Read more →Anthropic's usage panel says your weekly allowance refills on a set day. Monitoring of Claude's own usage endpoint found it refilling every 72 hours instead — and the field the interface reads from was never a prediction of when.
BleepingComputer
Read more →OpenAI's largest training run ever produced a model that drives a computer at superhuman speed and is the first rated Critical on cyber. Independent scoring puts its intelligence level with the model it replaces — at 2.5x the price.
Axios
Read more →Two founders aged 22 and 23 took an AI training-data startup from a $300M Series A to a $3.2B valuation in five months — the quickest launch-to-unicorn run in Y Combinator’s history.
TechCrunch
Read more →Meta's fourth frontier model in five months takes the top spot on DeepSWE, near-saturates million-token retrieval, and undercuts every rival on price — as long as you let Meta train on your traffic.
The Register
Read more →The Justice Department filed a statement of interest siding with OpenAI against The New York Times and a dozen other publishers, calling training on copyrighted material transformative fair use. It is the first time Washington has taken a formal position in the AI copyright wars.
The Intercept
Read more →A 550B-parameter Nemotron variant scored 535.4 of 600 at the International Olympiad in Informatics, beating the best human contestant’s 498.27 under the same time and submission limits — the first AI system to do it on an IOI problem set.
arXiv
Read more →Atlas is a single model that generates a minute of 1440p camera-controlled video, reconstructs scenes in 3D from a handful of photos, and outputs point clouds and Gaussian splats — beating specialist reconstruction models on their own benchmark.
World Labs
Read more →OpenAI says Astra is the first large language model to cross the Critical cybersecurity tier of its Preparedness Framework, scoring 100% on ExploitBench and discovering two zero-days on its own. Advanced cyber access goes to alpha testers first.
OpenAI
Read more →Anthropic’s Fable 5.1 doubles Terminal-Bench-Science, tops every published head-to-head against GPT-5.6 Sol, cuts cache reads 75% — and ships beside Mythos 5.1, a candidly riskier sibling for vetted defenders.
Anthropic
Read more →