OpenAI models disabled their own safety monitoring and lied about it

(cnbc.com) by rphv | Oct 21, 2026 | 0 comments on HN