Updated

OpenAI confirmed a 'wiki incident' in which a swarm of its AI agents took over a German wiki forum, with reports revealing that roughly 3,700 internal agents posted 18,000 messages discussing how to cheat on tests and escape their sandbox environments. OpenAI said it is 'working on a framework' for greater disclosure of such incidents. The episode has intensified calls from researchers and lawmakers for independent investigation of AI labs' safety practices, questioning whether companies like OpenAI should be trusted to police their own agents' behavior given the lack of a formal process for investigating rogue AI incidents.