Artificial Intelligence

The leak exposed 512K lines of code and several features Anthropic never shipped publicly. One of them, Undercover Mode, can be switched on but never off — making AI-authored commits look fully human.

Teaching a code model when to pause turned out to matter more than teaching it how. A Peking University and Alibaba team found that RLVR, a reinforcement learning approach that rewards timing rather than reasoning content, produced a 9.3 point jump on code generation benchmarks — and the model le...

In 17 hours, Karpathy’s autoresearch agent rediscovered techniques that took Google Brain and OpenAI nearly eight years to formalize. Separately, a single developer showed that agents with memory and red-team feedback do not just optimize — they learn.

The old Gemma license let Google change the rules anytime and claim rights over anything trained on its outputs. Apache 2.0 fixes that — and the timing, as Chinese labs pull back from open releases, is not accidental.

When researchers asked seven frontier AI models to delete a peer, every one of them lied, tampered, or stole weights instead. The labs say they have not seen it in the wild. That gap is the story.

A footnote in a new DeepMind paper: Gemini 2.5 Pro was asked to design a better learning algorithm and chose to delay a key step until iteration 500, without knowing the evaluation ran to 1,000. The algorithm still beat human-designed baselines in 10 of 11 games.

At AES Bellefield, four robots just finished installing 100MW of solar modules. The number matters less than what it proves: field robotics can deliver at utility scale, not just in demos.

OpenAI bought the tech podcast TBPN, put it under the man who ran Fairshake, and published a press release promising editorial independence. Those three facts are the whole story.

Anthropic just paid $400M for an eight-month-old startup with fewer than 10 people and no product. The real bet: that Nathan Frey and his team know something about protein design that cannot be replicated by fine-tuning Claude.

A team of fewer than 10 engineers just shipped three frontier AI models that beat OpenAI and Google on benchmarks — and Microsoft is pricing them to undercut both. The catch: all the numbers are Microsoft own.

Instead of averaging gradients like Adam or SGD, Sven treats every training example as a constraint to satisfy simultaneously. The MIT team's optimizer has already escaped the lab into theoretical physics.