HR 10567
To require entities to include human-controlled shutdown mechanisms in all artificial intelligence systems.
TL;DR
Rep. Thomas Kean (R-NJ) introduced a bill requiring all AI systems to include a human-controlled 'kill switch' that can shut them down. Any company that builds, deploys, or operates AI would need to prove a human can turn it off. The bill sits in the House Science Committee with no hearing scheduled yet.
How This Might Impact Your Business
AI developers and deployers across every sector would need to build or retrofit human-controlled shutdown mechanisms into their systems, including SaaS platforms, autonomous vehicles, industrial robots, and customer-facing chatbots.
Companies embedding third-party AI (like OpenAI, Anthropic, or Google models) into their products would likely inherit compliance obligations and need vendor attestations confirming shutdown capability.
Manufacturers of autonomous systems (drones, self-driving vehicles, warehouse robotics) face the biggest engineering lift since 'shutdown' for physical systems raises safety and liability questions.
No company size threshold is specified in the current text, meaning startups and Fortune 500s would face the same baseline requirement.
No specific penalties, enforcement agency, or effective date are detailed in the introduced version, so the compliance bar remains undefined until committee markup.
Financial services, healthcare, and HR tech firms using always-on AI (fraud detection, patient monitoring, resume screening) would need to document how a human operator can halt the system without disrupting critical operations.
Bill status is early: referred to the House Committee on Science, Space, and Technology with no hearing scheduled, so passage this session is unlikely.
What Should You Do
Ask your engineering or IT team to inventory which AI systems you build or use, and document whether each has a human override or shutdown capability today.
Send vendor questionnaires to your AI providers asking how their systems can be halted by a human operator and whether they would support customer-side shutdown controls.
Have legal and compliance track this bill through the House Science Committee; sign up for committee alerts and watch for a hearing announcement.
Brief your product and risk teams on 'human-in-the-loop' design principles now, since similar shutdown requirements appear in the EU AI Act and state-level proposals.
Assign one executive (CTO, Chief AI Officer, or GC) to own AI governance readiness so you can respond quickly if this bill or a similar one advances.
Who It Affects
Sponsors
Status Timeline
committee
Referred to the House Committee on Science, Space, and Technology.
September 24, 2026