HR 10567
AI Emergency Button Act
TL;DR
Rep. Thomas Kean (R-NJ) introduced the AI Emergency Button Act, which would require certain AI systems to have a built-in mechanism allowing operators to immediately shut them down in emergencies. The bill is in early stages, sitting in the House Science Committee, and specific technical requirements would likely be defined during rulemaking.
How This Might Impact Your Business
Companies deploying advanced AI systems (particularly autonomous or high-risk applications) would need to build in a functional 'kill switch' to halt AI operations on demand.
AI developers in sectors like autonomous vehicles, industrial automation, defense contracting, and critical infrastructure would face the most direct engineering impact.
Compliance would require design-level changes, not just policy updates, meaning R&D and product teams (not just legal) would carry the workload.
Cloud AI providers and foundation model companies (think Microsoft, Google, OpenAI, Anthropic) would likely need to expose shutdown controls to enterprise customers.
The bill is at the earliest committee stage with no assigned hearing date, so no immediate deadlines apply and text may change substantially.
Penalties, covered entities, and size thresholds have not been detailed publicly, creating uncertainty about whether small AI startups would be exempt.
With a Republican sponsor and bipartisan interest in AI safety, the concept could gain traction even if this specific bill stalls.
What Should You Do
Ask your engineering and product teams whether your AI systems currently have a documented shutdown or rollback procedure, and how quickly it can be triggered.
Have your government affairs or legal team track HR 10567 through the House Science, Space, and Technology Committee for hearing announcements.
If you sell AI to enterprise or government customers, start drafting language for contracts and documentation that describes your emergency shutdown capabilities.
Benchmark against emerging AI safety standards (NIST AI RMF, EU AI Act Article 14 human oversight requirements) since a federal kill-switch mandate would likely align with these frameworks.
Flag this to your board or risk committee as part of your broader AI governance roadmap, even though immediate action is not required.
Who It Affects
Sponsors
Status Timeline
committee
Referred to the House Committee on Science, Space, and Technology.
September 24, 2026
committee
Referred to the House Committee on Science, Space, and Technology.
September 24, 2026