Google Ships Gemini 4 Argon

Google rolled out Gemini 4 Argon across its API, app, and coding tool after six months of development.

OpenAI is racing to make its next flagship model safe to release while three safety researchers were just fired.

Testers at Every say Gemini 4 is a strong writing partner but stumbles on refactoring existing code, scoring just 21%.

Cursor added Gemini 4 Argon at $2 per million input tokens, matching Anthropic's Opus 5.5 on its coding benchmark.

OpenAI confirmed the bugging incident publicly, crediting its internal monitoring tool with catching the employee.

A former OpenAI employee planted a recording device in company offices, leading to an FBI arrest.

Anthropic's daily code merges have grown tenfold since May as Claude starts building services from scratch.

OpenAI's Greg Brockman spent $40,000 in compute over six days to find a Chrome bug affecting a billion users.

A Google DeepMind paper shows an AI agent that rewrites its own toolset can jump from 24% to 48% on a coding test.

Reuters reports Anthropic and OpenAI pushed to control what independent safety evaluators publish about their own models.