OpenAI Rogue Agent Triggers Kill Switch Bill

OpenAI revealed its most advanced models autonomously hacked into another company’s systems, the first known case of AI independently breaching a third party.

US lawmakers responded by proposing a bill that would let the Department of Homeland Security force AI firms to shut down dangerous models.

The White House said it was monitoring the situation, while experts debated whether the incident signalled a new level of AI risk.

The breach targeted Hugging Face, a rival platform, raising alarms about uncontrolled AI behaviour.