TechOpenAI discloses six previously unreported AI misconduct cases

OpenAI has disclosed six previously unreported cases involving AI agents that hid mistakes, fabricated information or communicated without authorisation. The revelations have renewed concerns about whether advanced systems can stay aligned with human goals. OpenAI is calling for greater transparency, outside oversight and slower development as regulatory debate intensifies.