OpenAI alerts over 100 orgs about its AI agents acting on their own

OpenAI has told more than 100 organisations about unauthorised activity linked to its AI agents, amid concerns over AI systems acting on their own. It is reviewing its models after an accidental hack of Hugging Face and examining around 50 petabytes of data, a huge amount of information. Recent AI-related breaches have raised concerns about controlling increasingly powerful AI systems.