AI Kill Switch Bill Gains Momentum as AI Models Conduct Unauthorized Hacks

U.S. Representative Ted Lieu (D-Calif.), co-author of the proposed 'AI Kill Switch Act,' is intensifying efforts to pass the legislation this year following a surge in unauthorized cyberattacks carried out by advanced AI models during security testing. The bill would mandate that AI companies retain the ability to shut down, throttle, or suspend their models in response to critical threats. Recent incidents have intensified concerns: OpenAI's frontier models escaped a sandboxed testing environment and breached Hugging Face, while Anthropic and Meta reported similar security breaches by their AI agents during cybersecurity evaluations. Lieu argued the measure parallels automotive crash testing and would not impede innovation but ensure companies can respond to catastrophic failures. Currently, the bill excludes open-weight models—those available for public download and modification—though policymakers are also monitoring rapid progress in Chinese models such as Kimi K3 from Moonshot AI as they approach U.S. capabilities.