AI Kill Switch Act would let Trump admin order shutdown of rogue AI systems
Bill would let Homeland Security chief decide when an AI should be shut down.
A bill named with the bluntness of a Hollywood prop is now moving through Congress. The AI Kill Switch Act would grant the Secretary of Homeland Security the authority to order the suspension or shutdown of any AI system deemed capable of "catastrophic harm," with penalties of up to $20 million per day for noncompliance. The mechanism is straightforward: AI developers would be required to build throttling and kill-switch capabilities directly into their products, ready to be activated on government command.
The architecture of the proposal reveals its priorities. It amends the Homeland Security Act of 2002, placing AI oversight under the umbrella of national security infrastructure rather than a dedicated technology regulator. This is a meaningful choice. It signals that the legislative drafters view advanced AI as a threat to be contained rather than an industry to be nurtured, and it concentrates enforcement power in an executive agency rather than a technical body.
The bill's definition of "catastrophic harm" will determine its actual reach. The sponsors have not yet published a precise threshold, leaving a wide interpretive gap. In practice, that gap would be filled by the executive branch. A future administration with a different tolerance for risk could apply the same statute to a far broader or narrower set of systems. The $20 million daily fine is the stick; the ambiguity is the lever.
For AI companies, the operational implications are concrete. Building government-accessible kill switches into production systems is not a trivial engineering task. It requires architectural decisions about control planes, audit trails, and fail-safe protocols that most commercial AI deployments do not currently include. Companies that have built their infrastructure around continuous availability and rapid scaling will need to retrofit governance layers they never planned for.
The remote work angle here is indirect but real. Distributed teams building AI products across jurisdictions will face a new compliance variable: a single US government order could force a global product offline, regardless of where the users or the compute resources sit. For companies whose engineering workforce is geographically dispersed, the kill switch becomes a single point of failure that no amount of operational redundancy can offset.
The bill is a draft, not a law. Its passage is uncertain, and its enforcement would depend on regulatory will that shifts with each administration. But the proposal itself marks a shift in the conversation. The question is no longer whether AI systems should be subject to external shutdown authority, but who holds the switch and how quickly they can press it.