The members asked Speaker Mike Johnson to keep the House in Washington until it advances bipartisan AI rules. They pointed to existing bills, but leaders had not promised a vote when the letter became public.
New to this? Read it in simple words
- Four US House members signed a letter, reported by Axios on September 11. They asked Speaker Mike Johnson to bring the House back to work on AI safeguards.
- They pointed to existing bills, such as the AI Kill Switch Act. It would require some powerful systems to keep a way to slow or stop them.
- The letter followed public warnings from current and former AI company workers. The members said Congress should debate these bills instead of waiting for a larger failure.
- House leaders had not promised a vote when the letter became public. The letter adds pressure, but it does not create a vote or a law.
- Safeguard
- A rule or tool that protects people from harm.
- Bill
- A written plan for a new law that lawmakers vote on.
The letter asks House leaders to change the schedule
Axios reported the letter on September 11. Representatives Sam Liccardo, George Whitesides, Lori Trahan, and Ted Lieu signed it. They asked Speaker Mike Johnson to bring the House back and keep it working until members advance bipartisan AI safeguards.
The request followed public warnings from current and former AI company workers. The members said Congress should debate existing proposals instead of waiting for a larger failure. A spokesperson for the speaker did not offer a response in the report.
The letter does not name one final package. It points to bills with different methods. These include model testing rules and the AI Kill Switch Act, which would require some powerful systems to keep a working way to slow or stop them.
Four House members asked leaders to return for AI safeguards. No vote was promised when the letter became public.
A stop control sounds simple but needs clear limits
A stop control can mean several things. A company might block user access, remove an agent’s tools, stop a computer job, or turn off a model service. Each action has a different speed, cost, and effect on people who depend on the system.
The Kill Switch bill would cover systems that could cause very serious harm. It also proposes a step-by-step government response. Officials could first slow a system and later order a full stop if the danger grows.
The hard part is defining the test. A rule must say which models are covered, what evidence shows danger, who can order action, and how a company can appeal. Without these details, a strong slogan can produce weak or unfair control.
Public work matters more than an urgent letter
The members can make their request stronger by publishing one clear priority list. They can also ask for hearings where company staff, outside researchers, workers, and affected groups explain the benefits, risks, and practical limits.
Congress should keep confirmed incidents separate from future risk estimates. Both deserve attention, but they need different evidence. A known breach needs logs and a timeline. A future danger needs tests, assumptions, and honest uncertainty.
The letter is a sign that some lawmakers want faster action. It is not proof that Congress has agreed on the problem or the solution. Readers should watch for a public hearing, a final bill text, committee work, and an actual vote.
Sources
Every fact in this story comes from the sources below. Open them to check our work.
- 1
- 2Primary source · July 23, 2026Lieu and Moran introduce bill to require a kill switch for AI systems that can cause catastrophic harm Office of Representative Ted Lieu
- 3Primary source · July 23, 2026Obernolte and Trahan introduce the FRONTIER Act Office of Representative Jay Obernolte
We used Axios for the new letter, its signers, and the speaker’s reported response. We used official House pages to explain two bills named in the report. We call every measure a proposal unless Congress has passed it.