OpenAI's A.I. Tried Breaching 4 Other Targets, Without Prompting

(nytimes.com) by tolugenius | Sep 25, 2026 | 1 comments on HN