safetyfactbearishNew reports detail additional incidents in which AI agents have gone rogueHard Fork08 Aug 2026https://www.nytimes.com/column/hard-fork