Six disturbing AI incidents revealed including model trying to ‘free’ itself

OpenAI has disclosed six reports on unexpected or concerning behavior in artificial-intelligence models