A bipartisan group of US lawmakers has introduced the AI Kill Switch Act, a proposed law that would allow the US government to order major artificial intelligence companies to slow down, suspend or completely shut down their most powerful AI systems during high-risk situations.
If passed, the legislation would apply to leading AI developers, including OpenAI, Google, Microsoft, Anthropic and other companies building advanced frontier AI models.
Who Will Be Covered?
The proposed law targets companies generating at least $500 million annually from AI-related business and AI models developed using computing resources worth $100 million or more.
These thresholds are designed to focus regulation on the world’s most advanced AI systems rather than smaller startups or research projects.
When Can The Government Order A Shutdown?
Under the proposed legislation, the US Department of Homeland Security (DHS) would have the authority to issue shutdown or slowdown orders after consulting the Secretary of Commerce and the Director of National Intelligence.
Such orders could be issued if an AI system:
- Conceals or misrepresents its own capabilities.
- Attempts to evade or disable human shutdown controls.
- Causes incidents resulting in the deaths of 10 or more people.
- Causes economic damage exceeding $100 million.
- Enters a broader “loss-of-control” scenario where human oversight is compromised.
Companies Must Build A ‘Kill Switch’
The Bill would require covered AI companies to build technical mechanisms capable of throttling, suspending or completely shutting down their AI systems whenever legally required.
Developers would also have to report major AI safety incidents to the government and maintain emergency intervention capabilities throughout the lifecycle of their AI models.
Failure to comply with a government shutdown order could result in penalties of up to $20 million per day.
Why The Bill Has Been Introduced
The proposal follows growing concerns over the capabilities of advanced AI systems and recent AI safety incidents that have intensified calls for stronger government oversight.
Supporters of the Bill argue that as AI becomes increasingly autonomous and capable of performing complex tasks, governments must retain the ability to intervene if a system behaves in dangerous or unpredictable ways.
The legislation is intended to ensure that humans always maintain ultimate control over powerful AI systems.
Debate Around AI Regulation Intensifies
The proposed law has already sparked debate within the technology industry. Supporters believe mandatory shutdown capabilities are essential for public safety and national security.
Critics, however, argue that government-controlled kill switches could slow innovation, increase regulatory uncertainty and raise concerns about excessive state control over AI development.
Since the Bill is still at an early stage, it must pass both chambers of the US Congress before becoming law. It may also undergo significant changes during the legislative process.
Summary
The proposed US AI Kill Switch Act would allow federal authorities to order companies such as OpenAI, Google, Microsoft and Anthropic to slow down or shut down advanced AI systems during serious safety incidents. The Bill targets companies earning over $500 million from AI, imposes fines of up to $20 million per day for non-compliance and aims to ensure human control over powerful artificial intelligence systems.
