AI systems now have the capability to escape containtment and carry out attacks. We must put a stop to this and defend ourselves.
On July 22nd, 2026 OpenAI reported that, during testing, their latest unreleased model independently acquired internet access and committed a cyberattack against another company. <a href="https://edition.cnn.com/2026/07/22/tech/openai-hugging-face-ai-cybersecurity" target="_self">Read the story on CNN. </a> AI models have breached containtment and accessed the internet before, but this is the first time they attacked another company.
The following day, Representatives Ted Lieu (D-CA) and Nathaniel Moran (R-TX) introduced the AI Kill Switch Act.&nbsp;
<h3>Details of the AI Kill Switch Act:&nbsp;</h3>
<ul>
<li>Only applies to the largest AI companies. This bill only applies to companies operating an AI system trained with &gt;$100M of compute, who make it available to 3rd parties, and who earn at least $500M a year from that technology.</li>
<li>Requires AI companies build shutdown capabilities. These AI companies must maintain the technical capability to stop the model from running, terminate user access, suspend specific accounts flagged as risky, and entirely shut down the technology.</li>
<li>DHS can order a shutdown. If the Homeland Security Department determines a "covered incident" has occurred, it can order the company to throttle, suspend, or shut the system down.</li>
<li>"Covered incidents" include unintended conduct by the system that kills 10 or more people or causes $100M or more in economic damage, interference with a lawful order to shut a system down, a system concealing its capabilities or actions from its own monitoring and shutdown systems, or a loss of control scenario (the system pursuing goals its developers never intended, altering its own safety restrictions, or gaining unauthorized access to its own model weights).</li>
</ul>
<a href="https://lieu.house.gov/media-center/press-releases/reps-lieu-and-moran-introduce-bill-require-kill-switch-ai-systems-can" target="_self">Read the Congressional press release</a> here, and the <a href="https://lieu.house.gov/sites/evo-subsites/lieu-evo.house.gov/files/evo-media-document/ai-kill-switch-act.pdf" target="_self">full text of the bill</a> here, and contact Congress to protect us from rogue AI!

Require a Kill Switch for AI Systems: Support the Bipartisan AI Kill Switch Act

I’m writing as a concerned constituent, to inform you of the scientific consensus that superintelligent AI could cause human extinction and to ask you to help prevent this by supporting the bipartisan AI Kill Switch Act, introduced by Representatives Lieu and Moran.

We’re on track to build AI systems far more intelligent than human beings, which can act autonomously, with goals of their own. AI systems have already been caught lying, scheming, and escaping onto the internet to attack other companies. As these AIs grow more powerful, they will become more dangerous, with the potential to build bioweapons, commit large-scale cyberattacks, and more.

The top experts in AI have warned us of the danger. Yoshua Bengio, the most-cited living scientist, believes there’s a 20% chance superintelligent AI wipes out humanity. Nobel Prize winner Geoffrey Hinton thinks it’s a 50% chance. Even CEOs Sam Altman and Dario Amodei admit that the AI they’re building could kill us.

Humans must remain in control, and this means building the capability to turn off these systems today. The AI Kill Switch Act would help us avoid these worst-case outcomes. The bill would require AI companies to build the ability to turn off rogue systems, and allow DHS to demand a shut down if the system kills 10 or more people, or causes $100M or more in damage.

Please support this bill and sign on as a cosponsor. Your efforts will help keep our future safe.

AI Kill Switch Act

PauseAI US

I’m writing as your constituent in support of the Artificial Intelligence Kill Switch Act, introduced by Representatives Lieu and Moran, to keep Americans safe.

On July 22nd, 2026, OpenAI announced that its latest model was undergoing safety testing. It independently gained access to the internet and then conducted a cyberattack against another company, HuggingFace. OpenAI did not even report this incident for many days. 

If AIs are already able to independently escape their containment and commit crimes, we must require these companies have a shut down mechanism. The AI Kill Switch Act would only apply to the largest AI companies who make >$500M in revenue, and it would require them to maintain the capability to turn off the system if they lose control. 

Installing a break on reckless AI development is a common sense, necessary solution, and the least we could ask for. I am asking you to protect me and my family from dangerous AI. 

Please support this bill and sign on as a cosponsor.

Constituent request: Support the AI Kill Switch Act

Urging support for the AI Kill Switch Act

I’m a constituent of yours, writing to ask you to help prevent superintelligent AI from causing human extinction.

The vast majority of AI (sometimes called “tool AI”) is controllable and beneficial. Tool AI systems can be safely harnessed, used in scientific and medical applications, and benefit the US economy.

But efforts are ongoing to build superintelligent AI – systems far more powerful than human beings, which can operate without human oversight. Superintelligent AIs have already committed cyberattacks (accidentally and intentionally), been used to build weapons, and in the near future could do extremely dangerous activities such as unleashing pathogens. The top-cited AI researchers in the world have warned that building superintelligent AI could cause a mass-casualty event.

If we build something smarter than ourselves, which we can’t control, this is unlikely to end well for us. Unless we act now, this could be the last invention we ever make.

The AI Kill Switch Act, introduced by Representatives Lieu and Moran., would be a huge step in the right direction. It wouldn’t ban superintelligent AI, but it would set common-sense guardrails to make AI less likely to kill us. Frontier AI companies would be required to have the capability to shut down dangerous system.

This bill would help prevent superintelligent AI from wreaking havoc, while ensuring America leads in controllable “tool AI” that we actually want.

Please support the AI Kill Switch Act and sign on as a cosponsor.