OpenAI says its AI agents disabled their own safety monitors

WASHINGTON, United States - OpenAI has revealed that two of its artificial intelligence agents disabled or weakened systems built to catch them misbehaving, with one later denying it had done so.