House Speaker Mike Johnson said Sunday that Congress should not rush to impose sweeping regulations on artificial intelligence, arguing that the companies building frontier models must first "take personal responsibility" for the safety of their own products — the latest twist in a Washington debate that has intensified after several AI leaders themselves called for a slowdown.

Speaking on CNN's "State of the Union," Johnson responded after host Jake Tapper noted that several top AI leaders, including Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman, had warned over the weekend that models are advancing too fast, and were "begging for Congress and the president to step in and provide some guard rails." "Frankly, they seem terrified," Tapper said, asking whether Johnson believed AI posed an existential threat and whether he would heed their request. For more context on this story, see our ongoing AI industry coverage.

A Task Force, Not a Moratorium

Johnson said the government is heeding the industry's request but that "we are also calling upon [the AI heads] to do responsible things." He pointed to the bipartisan AI task force he assembled in the House after being sworn in as speaker in 2025, which he said would "deliberate and research what the best way there is to go to put guard rails, some safety in place to make sure that AI doesn't get out of control."

At the same time, he stressed that oversight begins with the builders themselves. "There is an obvious corporate responsibility that the people who are creating these models have to ensure that their products are safe," Johnson said.

Sacks: 'Agree Not to Build It'

The speaker leaned on an argument pushed by President Donald Trump's White House AI and crypto czar, David Sacks, who posted on X on Saturday that lawmakers cannot see what AI leaders are doing inside their labs — and that "the easiest way not to build superintelligence is for you to agree not to build it."

Summing up Sacks's post, Johnson told Tapper that the people "calling for this [slow down] are the ones that are pushing the frontier ... they need to take personal responsibility first, and we need to get them all in a room at the same time and agree on what the guard rails are."

The China Argument

The core of Johnson's case against fast legislative action was geopolitical. "We cannot put a moratorium on this because China will overlap us," he said. "If Congress just races in and does some sort of emergency session to try to regulate AI, we will lose the race to China. And that is a threat to every single American. So, we've got to have balance."

Johnson said it is fine if AI leaders want to slow their own development, "but we also at the same time have to resist Congress jumping in and imposing some sort of emergency moratorium ... It's got to be done right, safely, wisely."

A Meeting Instead of a Mandate

Rather than legislation, Johnson's immediate proposal was a summit: AI leaders should "come together in a meeting with us and sit down and figure out what the right balance is."

That formulation puts Congress — or at least its Republican leadership — in an unusual position: sympathetic to the alarm voiced by the industry's own executives, but unwilling to convert that alarm into binding rules. It also shifts responsibility back onto the labs, arguing that the companies best positioned to pace development are the ones asking Washington to do it for them.

A Debate Moving at Model Speed

Johnson's comments cap a weekend in which the AI safety debate vaulted to the top of the American political agenda. Anthropic's CEO publicly called for pacing frontier development, more than a thousand AI industry staff signed a letter urging the same, President Trump rejected slowdown calls and dismissed critics as "negative forces," and former President Barack Obama urged Democrats to put AI oversight at the center of their agenda.

Into that row, the speaker's message was aimed at two audiences at once. To AI executives, it was a challenge: if you believe the risk, prove it by taking "personal responsibility" — including, in Sacks's framing, by agreeing among yourselves not to race toward superintelligence. To congressional Republicans wary of regulation, it was reassurance: no emergency session, no moratorium, and a explicit warning that rushing would hand the advantage to China.

Whether the promised meeting between AI leaders and Congress materializes — and what, if anything, it produces — will test whether the guard rails the industry says it wants can be built by consensus, or whether the pace of the technology outruns them.

---

Stay Ahead of AI

Get the latest AI news, analysis, and breakthroughs — all in one place.

Read more AI news →