US Lawmakers Propose 'AI Kill Switch' Act: What It Says and Why It's Controversial
A new bill would let Homeland Security shut down frontier AI systems—supporters say it's a safety net, critics call it overreach.
The proposed AI Kill Switch Act would give the Department of Homeland Security authority to order the immediate shutdown of any 'frontier' AI system deemed a national security risk, with daily fines up to $20 million for noncompliance. The bill has sparked debate over safety versus government overreach.
What the AI Kill Switch Act Proposes
Under the legislation, DHS could issue an emergency order to throttle or shut down any AI model that poses an 'imminent threat' to critical infrastructure, public safety, or national security. Noncompliant companies face fines starting at $1 million and escalating to $20 million per day. The bill targets so-called 'frontier' models—advanced AI systems with capabilities approaching or exceeding human-level performance in key domains.
The Case for the Kill Switch
Supporters, including several cybersecurity and AI safety groups, argue that as AI systems become more powerful, they could be used to automate cyberattacks, design weapons, or disrupt power grids. They say the kill switch is a last-resort safety mechanism similar to emergency brakes in other technologies. 'This isn't about stifling innovation; it's about having a circuit breaker when a system goes rogue,' one lawmaker stated in recent coverage.
The Case Against the Kill Switch
Critics—including free-market advocates, some tech companies, and civil liberties groups—warn that the bill grants unelected officials sweeping power over a fast-moving industry. They argue that the definition of 'imminent threat' is vague and could be used to target unpopular but lawful AI systems. Others note that many frontier models are open-source, making a shutdown order impractical. 'This could kill US AI leadership without meaningfully improving safety,' said a tech policy analyst.
Additionally, legal experts question whether DHS has the technical expertise to assess threats from complex AI systems, and whether fines of $20M/day are proportionate. The bill does not specify an appeals process, raising due process concerns.
What is the AI Kill Switch Act?
A proposed US law that would let the Department of Homeland Security order the shutdown of frontier AI systems posing an imminent national security threat, with fines up to $20 million per day for noncompliance.
Why are some people against the AI kill switch?
Critics say it gives government too much power over AI development, uses vague definitions, threatens innovation, and may be ineffective against open-source models.
What constitutes a 'frontier' AI model under the bill?
The bill targets advanced AI systems with capabilities approaching or exceeding human-level performance in areas like code generation, cybersecurity, or weapon design—though exact thresholds are not yet defined.
Has any AI system ever been shut down by the government?
Not under this law—it's still a proposal. No large-scale AI shutdown has been ordered by the US government to date.
What happens next for the AI Kill Switch Act?
The bill must pass through committee hearings and votes in both chambers of Congress. Experts recommend watching for markup sessions and expert testimony.