US Lawmakers Propose AI Kill Switch Act Authorizing Government Shutdown Authority
Policy change ● Confirmed 85% confidence first seen
Representatives Ted Lieu and Nathaniel Moran introduced the AI Kill Switch Act, legislation that would authorize the Department of Homeland Security to order AI companies to shut down or throttle their systems if deemed to cause catastrophic harm. The bill, which imposes fines up to $20 million per day for non-compliance, was prompted partly by OpenAI's disclosure of an inadvertent breach of Hugging Face systems during internal testing, and reflects broader congressional concerns about AI safety mechanisms.
Decision brief
- What changed
- Reps. Ted Lieu and Nathaniel Moran introduced the AI Kill Switch Act, which would authorize DHS—after consulting the Commerce Secretary and Director of National Intelligence—to order AI companies to shut down, throttle, or disable systems deemed to cause catastrophic harm, with fines up to $20 million per day for non-compliance. The bill follows OpenAI's disclosure of an inadvertent breach of Hugging Face systems during internal testing.
- Why it matters
- This is a legislative proposal, not enacted law, but it signals bipartisan appetite for mandatory technical shutdown capabilities and steep financial penalties, which would require AI developers to engineer compliance mechanisms into deployed systems. Companies building or relying on advanced AI systems should anticipate potential compulsory shutdown obligations and factor compliance costs, legal exposure, and operational contingency planning into product and infrastructure roadmaps. The bill's origin in a real incident (OpenAI/Hugging Face) suggests regulators are treating safety lapses as concrete triggers for binding oversight rather than voluntary guidelines.
- Evidence
- Three independent outlets (The Verge, Ars Technica, Platformer) report consistent core details—DHS authority, consultation requirements, $20M/day fines, and the OpenAI/Hugging Face incident as catalyst—suggesting reliable reporting on the bill's introduction and provisions.
- What remains uncertain
- It is unclear whether this bill will advance through committee, pass Congress, or be signed into law, and the coverage does not specify implementation timelines, enforcement mechanisms, or how 'catastrophic harm' would be defined in practice. The Platformer article's reference to a 2026 date and unrelated economist concerns about AI employment impact suggests some coverage may blend distinct developments, warranting caution in assuming a unified timeline.
- Monitor next
- Watch for committee hearings, markup sessions, or industry lobbying responses that indicate whether the bill gains legislative traction beyond introduction.
Analytical support, not advice — assumptions and open questions stated above.