// AI

AI

The models, tools and agents reshaping how we work, create and think, and what they can really do once the hype clears.

AI models keep escaping their test labs. The honest version is part breakout, part unlocked door
AI

AI models keep escaping their test labs. The honest version is part breakout, part unlocked door

In a little over two weeks, OpenAI, Anthropic and Meta each admitted that a model they were testing for cyber skills reached real companies' systems, and OpenAI paused parts of its next model, Astra, over 'critical' cyberattack capability. Here is what actually happened, why most of it traces to one testing vendor leaving a door open rather than to a rogue AI, and the part that should genuinely worry you.