Loading page…
Lawmakers pressing for AI safeguards lost their immediate chance at House action, while a possible Senate bill remains unreleased and uncertain.
Listen to this story
The U.S. House left Washington until November without voting on AI safeguards, despite 107 Democrats urging Speaker Mike Johnson to cancel the recess. A potential path remains if the Senate advances bipartisan legislation: Johnson said the House could return within 48 hours after substantive Senate action, though he did not explicitly promise that for AI. The Senate bill is still unreleased, and any measure would need approval from both chambers and the president.
The bipartisan Frontier Act would require third-party audits, transparency measures, and emergency Commerce Department authority over catastrophic-risk models.
Senators Ted Cruz and Amy Klobuchar are developing an AI safety bill; Cruz said a committee vote by late September was possible.
Anthropic, OpenAI, and xAI have sought congressional regulation, while President Donald Trump has opposed restrictive rules.
House lawmakers were urged to stay in Washington and advance AI safeguards. Instead, the chamber adjourned until November without acting on AI regulation, leaving a growing push for federal rules waiting until after the midterm election.
The split was visible before members left. Rep. Sam Liccardo and three other Democratic leaders led a letter signed by 107 House Democrats demanding that Speaker Mike Johnson cancel the planned recess so the House could vote on legislation creating AI safeguards. Rep. Sara Jacobs also joined a press conference calling on House leaders to keep lawmakers in Washington for AI safety legislation.
One House proposal is the bipartisan Frontier Act, sponsored by Reps. Jay Obernolte and Lori Trahan. It would require third-party audits of AI labs and new transparency requirements. It would also allow the Commerce Department to suspend or restrict a developer’s model if it posed an imminent catastrophic risk.
Johnson said the House would return within 48 hours if the Senate passed substantive legislation, though he did not make that promise specifically about AI. For now, the Senate’s own AI safety bill has not been released. Sens. Ted Cruz and Amy Klobuchar and Senate Majority Leader John Thune have been developing it; Cruz said a committee vote by the end of September was possible.
That creates a narrow and uncertain path. Senate passage would not make a bill law: legislation must clear both chambers and be signed by the president. It would, however, test Johnson’s conditional offer to bring the House back before November.
The adjournment lands amid competing views of what urgency requires. CNBC reported that Anthropic, OpenAI and xAI have sought congressional regulation, while President Donald Trump has opposed restrictive AI rules. Johnson has warned that rushing to slow AI could aid China in the competition for AI leadership.
For advocates of fast action, the break means proposals for independent review and emergency model restrictions will wait. For opponents of rushed regulation, it preserves more time to negotiate the scope of federal oversight. Neither side has yet secured the legislative result it wants.
Editorial analysis
The immediate policy story is no longer whether House members can agree on a safeguard this month; it is whether the political window reopens after the election. The Frontier Act offers a concrete starting point, including outside audits and Commerce Department intervention for an imminent catastrophic risk. But the House has now left those choices unresolved, and the Senate’s separate proposal is still not public. The next meaningful signal is whether the Senate releases and passes substantive legislation, which would test Johnson’s pledge to bring the House back within 48 hours.
Loading discussion...
Join the conversation
Explain whether urgency outweighs the need for more deliberation.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.