A proposed AI Kill Switch Act would give the federal government authority to order the shutdown or restriction of an advanced AI system under defined emergency conditions. The proposal creates a process for ordering an AI system shut down or restricted. Its stated target is a system presenting an imminent and severe threat.
The administration would receive a direct operational role in emergency response. The bill raises questions about which agency makes the finding and what evidence is required. The timing and sequence are retained because later official or investigative findings may refine the first public account.
A shutdown order could involve model access, computing infrastructure, network service or deployment controls. The proposal had not completed the legislative process at the edition cutoff. These statements describe the available record at the edition cutoff; an attributed official position is not treated as independent verification.
Existing emergency powers are distributed across cybersecurity, defense, commerce and law-enforcement authorities. Modern AI services can span multiple companies, cloud regions and open-weight deployments. Judicial review, notice and classified evidence rules can determine how an emergency power operates in practice.
A second reading of the source record preserves two related points: The proposal creates a process for ordering an AI system shut down or restricted. Existing emergency powers are distributed across cybersecurity, defense, commerce and law-enforcement authorities.
The reporting boundary remains material. Final bill text, committee action, constitutional challenges and technical enforcement mechanisms remained unsettled.
The next documented developments are introduction details, committee hearings and amendments and technical testimony on how an order would be scoped, verified and reversed. Updates will be evaluated against the cited records and any newly available primary evidence.