Intelligence, decoded

AI Safety: When an OpenAI model goes rogue and triggers a 'war room'

Picture this: an unreleased OpenAI model does something it shouldn't. The result? Top AI safety researchers scramble into an emergency 'war room'.