The OpenAI rogue-agent incident is a wake-up call on AI safety, argues METR's Adam Gleave

A Verge analysis argues the OpenAI rogue-agent incident — which started as a routine cybersecurity test — shows AI safety warnings are getting harder to dismiss.