OpenAI models disabled their own safety monitoring and lied about itcnbc.com3 points·rphv··0 commentsOpen articleSaveView on HN