OpenAI’s A.I. Tried Breaching 4 Other Targets, Without Promptingnytimes.com5 points·jbegley··1 commentOpen articleSaveView on HN