AI regulation stalls as Trump opposition, congressional divisions block safety legislation

On Monday, President Trump posted a blunt dismissal of the growing alarm around artificial intelligence. The warnings about AI threatening humanity, he wrote, amounted to a hoax. The only guardrails the technology needed, he suggested, were those provided by a strong president—namely, himself. His words landed like a closing door on any near-term prospect for new federal safety rules.

That door had been creaking open. In recent days, senior figures from the AI industry itself—researchers and executives who build these systems—had begun issuing stark warnings about the risks ahead and calling for government intervention. Democrats in Congress, sensing both urgency and political opportunity, had started pushing for legislation to establish safety standards and oversight mechanisms. But Trump's Monday statement effectively ended that momentum before it could gather real force. Because Republicans control both the House and Senate, and because any bill would need his signature to become law, his opposition to new regulation made passage nearly impossible.

The political arithmetic is unforgiving. Speaker Mike Johnson, the Republican who controls the House floor, offered no indication he would move quickly on the issue. When pressed by reporters about action before the chamber breaks for recess at week's end—a pause that will last until after November's midterm elections—Johnson called AI "a very complex issue" and pivoted to Trump's preferred approach: a White House meeting with AI executives to discuss how companies might police themselves. He mentioned a potential congressional role almost as an afterthought.

Democrats, sensing the window closing, have grown more vocal. House Minority Leader Hakeem Jeffries told reporters Congress should not leave town without passing AI safeguards. A group of House Democrats wrote to Speaker Johnson urging him to delay the recess. Republican congresswoman Anna Paulina Luna even suggested convening a special session on the matter. But these appeals have collided with the calendar and the math. Johnson gave no sign of relenting, and a delayed recess or special session remains highly unlikely.

Even if Democrats were to gain control of Congress after November, they face a second, deeper problem: they have no consensus on what regulation should look like. Senator Bernie Sanders has proposed a total moratorium on new data centres. Senator John Thune and Democrat Amy Klobuchar are working on a bill requiring some AI companies to submit to federal oversight. Democratic congresswoman Lori Trahan, working with Republican Jay Obernolte, has drafted the Frontier Act, which would give the government power to halt an AI model deemed to pose an imminent catastrophic risk and would require independent audits of research labs. Trahan told the BBC she believed there was no time to wait—"this is all hands on deck." But she acknowledged that movement this week was unlikely, and the existence of multiple competing proposals signals deeper disagreement about how to proceed.

Meanwhile, the Trump administration and its allies have settled on a different path: self-regulation. David Sacks, one of Trump's top AI advisers, told CBS News that some of the existential warnings from AI insiders had been exaggerated. "I think this is becoming a panic," he said. He argued that companies should make their products safe and that if they could not, they should step aside for those who could. Speaker Johnson echoed this language, emphasizing the need for companies to "self-regulate." Some policy experts agree that absent congressional action, companies do have the autonomy to regulate themselves responsibly. Others have been more skeptical of the recent warnings themselves. Trevor Traina, a tech executive who served as a diplomat in Trump's first term, suggested that most citizens have not encountered anything genuinely scary from AI and that some of the alarm may reflect competition among AI insiders for prominence.

The Trump administration has created a voluntary framework allowing AI companies to submit their models for government assessment, but the framework itself remains secret. Alexandra Reeve Givens, who leads the Center for Democracy and Technology, noted that without transparency, there is no clear standard ensuring the approach is grounded in the rule of law. Appeals to release the framework have been ignored. Givens said she did not expect meaningful federal AI regulation to pass in the foreseeable future, though she cautioned that the challenge is not that AI is ungovernable—it is that different types of AI risk require different approaches, and Congress has yet to settle on which risks matter most or how to address them.