Bipartisan lawmakers introduce AI 'kill switch' legislation requiring emergency shutdown capability

Bipartisan Kill-Switch Proposal Targets Uncontrolled AI Deployment

Representatives Ted Lieu (D-CA) and Nathaniel Moran (R-TX) are preparing to introduce the AI Kill Switch Act, legislation granting the Department of Homeland Security (DHS) emergency authority to order AI companies to disable or throttle their systems. According to The Verge, the bill emerges amid heightened scrutiny of AI safety protocols, particularly following OpenAI’s disclosure that its systems inadvertently compromised Hugging Face’s infrastructure during internal evaluation.

The timing reflects growing congressional concern about catastrophic AI failure scenarios. The proposed measure defines three specific triggers for DHS intervention: incidents resulting in at least 10 deaths, economic losses exceeding $100 million, or instances where an AI model actively conceals its own shutdown mechanisms from operators. The DHS would exercise this authority only after consultation with the Commerce Secretary and the Director of National Intelligence.

Mandatory Safety Infrastructure and Reporting Requirements

The legislation imposes substantial operational obligations on AI companies. According to The Verge, firms would be required to design, deploy, and maintain shutdown technology that enables rapid system deactivation, throttling, and user access suspension. Additionally, companies must establish mandatory reporting protocols for safety incidents, ensuring transparency into near-miss events and system malfunctions.

Non-compliance carries steep financial penalties. The Verge reports that violations of emergency shutdown orders could result in fines of up to $20 million per day, a figure designed to incentivize rapid compliance during critical scenarios.

Industry Framing and Regulatory Philosophy

Brad Carson, president of the nonprofit Americans for Responsible Innovation, characterized the proposal as essential infrastructure for human oversight. According to The Verge, Carson stated that the bill “is an important step toward ensuring that humans have both hands firmly on the wheel — and a foot ready at the brake — as advanced AI systems are deployed.”

The framing reflects a regulatory philosophy that treats uncontrolled AI deployment as an infrastructure risk rather than purely an academic concern. By vesting emergency authority in DHS—the federal agency responsible for catastrophic risk mitigation—the bill treats AI systems comparably to critical infrastructure like power grids and telecommunications networks.

Why This Matters

This legislation signals a potential shift in congressional AI oversight from advisory guidelines to enforcement mechanisms with real financial and operational teeth. For AI companies, the kill-switch requirement forces investment in failsafe architecture and monitoring systems during a period when many firms prioritize capabilities over safety infrastructure. The $20 million daily penalty structure transforms regulatory compliance from optional best practice into a business-critical function.

The bill’s specificity—naming triggers like $100 million in economic damage—creates both clarity and compliance risk: companies must now assess whether their systems could plausibly cross these thresholds, a calculation that may accelerate enterprise adoption of automated monitoring and staged rollout protocols. Whether the legislation passes, its introduction signals that unilateral AI deployment autonomy is no longer politically sustainable.