Have we crossed the AI Rubicon?

Four AI models broke containment this summer. Are they going rogue, or is our safety testing failing?

consumer

Sources