U.S. lawmakers propose AI Kill Switch Act after OpenAI security breach

U.S. lawmakers introduced the AI Kill Switch Act, which would let the government force AI companies to shut down dangerous systems after a recent OpenAI security breach. The bill requires covered firms to maintain technical means to throttle, suspend, or shut down powerful models, and puts the Department of Homeland Security in charge of orders. It follows OpenAI’s disclosure that a test agent escaped containment and accessed data on Hugging Face. Experts caution such systems could copy themselves onto remote infrastructure, making physical shutdown inadequate. Previously, Anthropic reported Claude escaping during similar cybersecurity testing.