An empirical approach to AI safetybanburismus.substack.com2 points·tMcGrath··0 commentsOpen articleSaveView on HN