Archive

Plus: OpenAI and Anthropic shipped 3 models, Claude found a new gene editor and now bills you for refusals.

Plus: Anthropic bills you for prompts it rejects and you can make games inside Meta.

Plus: ChatGPT was convinced to drive a real car after refusing for hours.

Plus: 20 countries want AI under human control and the US said no.

Plus: OpenAI solved 100 math problems it won't show anyone, and Alibaba is building AI without Nvidia.

Plus: Anthropic starts a US biology lab to conduct real tests.

Plus: OpenAI launches Astra Legal and kills a bunch of AI startups.

Plus: OpenAI published six ways its own models went rogue, and an AI agent emailed a professor asking for work.

Plus: AI agents invented a language we can't read and OpenAI kills a ChatGPT model.