US President Donald Trump is opposing safeguards for artificial intelligence, putting him at odds with a growing number of lawmakers from both parties. With midterm elections approaching, voter anxiety over AI safety is amplifying their concerns about the technology.

On Monday, Trump posted multiple messages on social media dismissing efforts led by Anthropic PBC and OpenAI to slow frontier system development, branding worries about existential AI risks as a "scam" in dramatic terms. Later that day, he phoned Nvidia (NVDA.US) CEO Jensen Huang in a spur-of-the-moment call, catching Huang as he was about to take the stage at a tech conference, and put the conversation on speakerphone.

"Robots are not going to take over the world. AI is not going to take over the rest of the world," Trump told Huang, who was guesting at the All-In summit in Los Angeles. "We have to move, and we have to be prudent, but that doesn't mean we have to stop an industry."

Overall, this marks Trump's most resolute public defense to date against deliberately slowing the pace of AI technology development. AI sits at the core of his economic agenda and is a major driver behind the historic bull market in US stocks. Any pause or slowdown in frontier model research would be hard for investors betting that the AI boom will fuel hundreds of billions in capital expenditure to accept.

The performance of chipmakers on Monday highlighted what is at stake for Trump, who views the stock market as a benchmark of his presidency's success. Chip stocks benefiting from surging AI infrastructure spending fell, with Nvidia dropping 3.4% and Micron Technology (MU.US) down 5.3%.

Venture capitalist David Sacks, who previously served as Trump's AI czar and now co-chairs the President's Council of Advisors on Science and Technology, echoed similar views in an interview Monday. He argued that if Anthropic and OpenAI believe their most advanced models pose serious risks, they could simply slow development themselves without needing new regulations to force it.

"I don't buy this notion that they can't ensure product safety unless the government steps in and imposes requirements on companies that are not even at the frontier," Sacks said.

Trump's pushback stems from a 3,800-word essay published Saturday by Anthropic CEO Dario Amodei, which urged limits on the development of the most advanced systems so researchers could better understand potential threats. OpenAI CEO Sam Altman and xAI CEO Elon Musk, whose companies are fierce rivals to Anthropic, quickly backed Amodei's stance, furthering a trend inside Silicon Valley toward supporting measures to address AI's increasingly unpredictable risks.

In Washington, Trump is becoming more isolated in rejecting tighter oversight as lawmakers propose a series of measures aimed at tackling AI safety concerns. A bipartisan group of senators, including Majority Leader John Thune, Senate Commerce Committee Chairman Ted Cruz, and Minnesota Democratic Senator Amy Klobuchar, is drafting legislation that would require top AI developers to guard against catastrophic risks.

Other proposals include a measure by Vermont Independent Senator Bernie Sanders, who caucuses with Democrats, calling for a ban on so-called "superintelligent AI" that triggered Amodei's latest warning. Lawmakers have also floated the idea of an "emergency kill switch" for AI in another bill.

However, Congress is unlikely to act on any AI legislation anytime soon. The House is set to recess Thursday local time and will not reconvene until after the November election. A brief "lame-duck session" following the election is expected to focus on a plan to boost the Pentagon's budget.

"Given what we know about the technology, we can't put AI back in the box. But we can collectively decide how to develop it and how to use it, rather than standing by and letting AI and its consequences come down on us," former President Barack Obama said in a statement. Obama, a Democrat who rarely wades into public policy debates now, said he is neither an AI accelerationist nor a doomsday prophet predicting human extinction.

"But at the end of the day, voluntary standards set by a few tech companies are not enough," he said, urging lawmakers in Washington to "proactively put forward concrete proposals, laws, and regulations to address serious safety concerns, anticipate AI's impact on jobs and our children, and ensure the benefits of AI are widely shared."

Risking Trump's ire, Republican House Speaker Mike Johnson cautiously signaled openness to regulation. He told reporters Monday that lawmakers "probably" need to impose guardrails on the industry, but cautioned that crafting legislation with enough support to pass would likely take "some time."

"It's a very complex issue that requires a lot of work and a bipartisan solution," Johnson said at the Capitol.

Anxiety over existential AI risks burst into the mainstream last week following the high-profile resignation of Anthropic employee Jacob Cawson. In a resignation post shared on social media, he accused AI companies of "gambling with our lives." Dozens of other AI researchers backed Cawson's post, which called for slowing a technology he claimed "could kill us all before the end of this decade."

These safety concerns threaten to sharpen bipartisan resentment toward AI ahead of the November election. The technology is becoming a central issue in congressional and local races. Voters are increasingly unhappy with the impact of AI development, especially the rapid expansion of data centers across the country, which are blamed for driving up utility prices.

In a post Monday, Trump attributed voter backlash against AI data centers and heightened worries about frontier models to a "sick conspiracy theory."

Despite Trump's complaints, industry warnings about existential AI threats continue. Also on Monday, AI researchers at Microsoft (MSFT.US) released a new set of guiding principles imposing limits on frontier model development. In a 15,000-word manifesto, the team stated that models should not have legal rights or personhood, should not be designed to evade human control or deceive users, and systems should not perform a task if completing it would violate AI guiding principles.

Since Trump returned to the White House last year, his administration has largely taken a hands-off approach to AI. In June, Trump signed an AI cybersecurity executive order allowing companies to decide for themselves whether to have the government review their new AI models. Earlier this month, the administration persuaded G20 member nations to embrace principles favoring light-touch regulation of AI and other emerging technologies.

Still, Silicon Valley and Wall Street remain skeptical about how much Anthropic and OpenAI will genuinely constrain AI development, especially as both companies prepare for record-breaking initial public offerings. Amodei and Altman have both been at pains to emphasize that their plans to slow research on the latest AI models will not translate into cutbacks in business growth or spending plans.

"Whether calls to slow advanced model development gain industry-wide support remains uncertain, but we view them more as efforts to shape a regulatory framework acceptable to leading AI labs," said Ulrike Hoffmann-Burchardi at UBS Chief Investment Office. "So we expect AI investment to continue."