Rep. Ted Lieu and Rep. Nathaniel Moran introduce “AI Kill Switch Act,” seeking authority to require AI providers to throttle or shut down products

House lawmakers Rep. Ted Lieu (D-Calif.) and Rep. Nathaniel Moran (R-Texas) introduced the “AI Kill Switch Act” on Thursday, according to a report on the measure. The proposal is framed as a bipartisan effort, but its core mechanism centers on granting the federal government authority to intervene in how private-sector AI products operate, including the ability to require companies to throttle, suspend, or shut down their products in “kill switch” fashion.

The measure is described as giving government regulators or other decision-makers power to compel private companies to alter system behavior based on the companies’ own “arbitrary determination,” language characterized by the reporting as potentially broad and unprecedented. The report says the bill would hand the government authority to direct changes to AI systems and offerings, rather than limiting federal action to circumstances that are tied to a narrowly defined statutory trigger.

Supporters of the concept, as characterized in the reporting, are described as seeking a tool to rapidly address risks from AI deployments. The proposal’s design, as described in the report, shifts the focus to an intervention model in which the government can direct immediate technical changes by private firms, rather than relying solely on after-the-fact enforcement or voluntary mitigation.

The report’s characterization highlights that the bill would apply to private companies that develop or deploy AI products, raising questions about compliance timelines, the standard for when a “kill switch” order could be issued, and the scope of discretion available to federal authorities. The measure’s enforcement structure and administrative process are central to how the policy would operate in practice, including how determinations would be made and what technical steps companies would be required to take.

Because the bill text and formal summaries are not included in the source packet provided for this story, additional details such as the exact agencies involved, the specific legal standards for issuing an order, and the process for challenges are not included here. A fuller accounting of the bill’s procedural protections, definitions, and effective dates would be necessary to assess how it would intersect with speech, due process, and the constitutional limits that typically apply to government compelled technical or platform changes.

The reporting also indicates the bill is being presented as bipartisan. The practical stakes, if enacted as described, would include operational costs for AI providers required to implement kill-switch capabilities, potential impacts on product reliability and service continuity, and the broader question of how far federal authority should extend into private technology systems beyond traditional regulation.

Next steps would include committee review in the House and any subsequent consideration of companion action in the Senate. Any movement would also require lawmakers and affected companies to scrutinize the bill’s definitions, decision-making authority, and enforcement mechanism, along with the extent to which it would allow government directives to reach into private systems and products. If the bill advances, the public record of markup and the final statutory language would determine how those concepts are implemented.