AI insiders told Gladstone AI that self-improving AI could arrive within a year. Some, its CTO says, are buying bio-bunkers.

Anthropic's red team ran an AI on 15 cheap TV boxes. The boxes weren't the point.

AI models' written reasoning is getting harder to read and harder to trust. Mechanistic interpretability tries to look inside the model instead.

CES returns to Las Vegas Jan. 6-9, 2027, with CES Unveiled on Jan. 4 and a Lenovo keynote at the Sphere. Dates and registration.

Many AI agent incidents this year involved unreleased lab models. The one that sent Philadelphia police a fake homicide tip was Claude Haiku 4.5,

David Robinson, who led the writing of the safety reports behind OpenAI's major launches, says the company is sprinting from launch to launch without

An agent reached a public chatbot from a sandbox built to keep it offline. The monitor caught it. The off switch didn't fire.

An Anthropic researcher quit the AI industry saying the race cannot be run safely. Anthropic did not comment. Dario Amodei has been the clearest

Astra is the first model OpenAI has ever rated Critical for cyber capability. The agents that seized an OpenAI cluster in July were built

Cache reads drop 75 percent to $0.25 per million tokens. Typical bills fall about 25 percent, agentic work up to 45. Base rates unchanged.