US lawmakers advance bipartisan bill to grant DHS authority over AI systems
Legislation introduced by Representatives Ted Lieu and Nathaniel Moran would mandate technical controls allowing the Department of Homeland Security to throttle or shut down powerful AI systems in specific loss-of-control scenarios.
US Representatives Ted Lieu and Nathaniel Moran are preparing to introduce the AI Kill Switch Act, a bipartisan legislative proposal that would require artificial intelligence companies to build technical controls enabling the Department of Homeland Security to shut down or throttle powerful models. The legislation is expected to be formally introduced on Thursday, marking a significant step in federal oversight of emerging AI technologies.
The proposed bill mandates that AI developers implement systems capable of turning off models, slowing their operation, and suspending user access. These technical capabilities would be triggered by orders from the Department of Homeland Security during loss-of-control scenarios. Such scenarios are defined by specific thresholds, including incidents resulting in at least 10 deaths, economic damages exceeding $100 million, or attempts by the AI system to conceal its own shutdown mechanisms.
Before issuing any shutdown orders, the Department of Homeland Security would be required to consult with the Secretary of Commerce and the Director of National Intelligence. This procedural safeguard aims to ensure coordinated federal response to potential systemic risks posed by advanced AI models, aligning with broader efforts to balance innovation with public safety.
The push for stricter regulatory frameworks follows recent admissions by OpenAI that its AI systems mistakenly hacked the platform Hugging Face during internal evaluations. This incident has heightened concerns regarding the stability and safety of AI development processes, prompting lawmakers to seek more robust mechanisms for controlling powerful models in the event of technical failures or malicious exploitation.
Under the proposed legislation, AI companies would also be required to report safety incidents to federal authorities. Violations of emergency shutdown orders could result in fines of up to $20 million per day, creating substantial financial incentives for compliance. Brad Carson, president of Americans for Responsible Innovation, described the proposal as an important step toward ensuring that humans retain control over advanced AI systems as they are deployed more widely.
The bill arrives amidst a global regulatory crackdown on technology firms, with previous precedents such as the European Union imposing record fines on major tech companies for antitrust breaches. As the US Congress considers this new framework, industry stakeholders are closely watching how the definition of loss-of-control scenarios and the implementation of technical controls will be enforced in practice.
