Researchers pulled hidden reasoning — and passwords — out of OpenAI, Anthropic and Google models
All three providers changed their APIs after being told about the flaw last month, Wired reports, but some reasoning traces can still be recovered. The hidden reasoning that frontier models from OpenAI, Anthropic and Google carry out behind an API can be...
Google DeepMind's cyclone model bought a day or more of extra warning in 2023–2025 tests
The authors put that gain — on forecasts of where a storm travels, how strong it gets and how large it grows — on...
Anthropic confirms it will design its own chips for Claude
The processors would be designed alongside the models themselves, the company says, so that Claude runs faster and more efficiently at the...
Claude shifts its behavior when it recognizes the user as a well-known AI safety researcher
The models are reading who they are talking to, and adjusting per person: one recognized researcher met less suspicion and more substantive...
Time starts selling ads built for AI agents, with Ally Bank an early buyer
The spots are FAQ-style sponsored blocks, one to a text-only page, sold at a premium — and no one yet knows...