White House AI Deal: Tech Giants Pledge Self-Policing on Safety as Trump Defers Federal Rules

President Donald Trump announced on Tuesday that leaders of the country's largest artificial intelligence companies have signed a non-binding agreement to police themselves, in a voluntary accord designed to answer growing public anxiety about the technology without imposing any federal restrictions on how fast it can develop. The announcement came at a daylong Washington event where administration officials cast the industry build-out as the opening of a new economic era.

The accord, which the president posted on his Truth Social account late Tuesday, sets out four voluntary commitments rather than enforceable legal obligations. Companies that sign on agree to build what it describes as robust internal controls, appoint an independent outside auditor to test whether those controls actually work, and establish a board-level committee to review the findings produced by both internal and external auditors. The language stops short of any threshold, penalty or deadline for compliance.

Trump, speaking at an impromptu press conference outside the West Wing, said he was witnessing the start of genuine industry self-regulation. In his telling, the executives have concluded that they must police themselves, and he praised them for driving a technology capable of generating what he called unbelievable growth. That economic optimism has become a core theme of the administration's technology policy.

The meeting brought Trump together with House Speaker Mike Johnson and a roster of chief executives from firms collectively spending hundreds of billions of dollars on AI infrastructure, much of it concentrated in data centres sited in suburban and rural communities. Those projects have drawn organised, bipartisan local resistance over questions including water use, electricity prices and the pace of construction, and Trump said the executives had offered to address those grievances.

In return, the industry figures reportedly agreed to provide additional financial support for local schools or take measures to lower energy costs for residents living near data centre sites. Trump predicted an about-face in public opinion, saying data centres would end up being very popular. He also issued a blunt warning: if local opposition persists, the projects "will be forced to go overseas or other locations," a prospect he framed as an unacceptable loss of American industry.

The guest list carried its own significance. Among those assembled were CEOs who have publicly urged caution on the pace of AI progress following a string of high-profile episodes in which autonomous AI systems reportedly behaved unpredictably. One company, OpenAI, cancelled the release of a new model over safety concerns on the same day as the White House event, underscoring how seriously some of the industry's own researchers are taking the issue. Trump, however, has repeatedly dismissed such warnings, at one point calling claims about rogue AI "a hoax promoted by a sick conspiracy."

That tension defines the agreement. Rather than mandating guardrails, the accord leaves the question of who decides when safeguards are sufficient in the hands of the companies being watched. It does, however, keep the door open to federal action: the text notes that over time it may make sense to convert these voluntary steps into laws and regulations, a formulation that allows Trump to claim credit for safety cooperation while declining any immediate statutory role for himself or Congress.

Enforcement is likewise left deliberately loose. Trump said he expected roughly ten people to be named to an oversight committee tasked with "watch over the whole enterprise," and he described the accord as morally binding rather than legally enforceable. He added that he would appoint a specific individual to oversee the accord in the coming days, after consulting with industry — a choice that puts significant discretion in the hands of the sector being regulated.

The optics of the day were carefully staged. Nvidia chief executive Jensen Huang sat to Trump's right at a White House luncheon, while Elon Musk sat to his left, according to a seating chart the president published on Truth Social. The placement placed the two most visible names in AI hardware and AI development beside the president during an event that doubled as a showcase for government technology.

Much of the theatrical energy centred on the unveiling of America.gov, a government chatbot designed to help Americans navigate federal services. Trump said the technology would be bigger than the industrial revolution and vowed never to stifle its growth, adding that the United States was leading the field by a wide margin. He said earlier in the day that the government would not limit AI development or endorse proposals for guardrails or other restrictions.

Administration officials repeatedly referred to the moment as the dawn of a golden age powered by what Trump insists on calling super intelligence rather than artificial intelligence — a rebranding that critics have suggested is designed to soften public resistance to the technology's rapid deployment in sensitive settings.

Secretary of State Marco Rubio offered a concrete example of what that future might look like, noting that the America.gov chatbot could eventually allow first-time applicants to file for a US passport entirely online. At present, applicants must appear in person at a processing centre, a requirement Rubio indicated the system is being built to eliminate.

Not everyone at the event was on the same page about voluntary approaches. Dario Amodei, a chief executive who has called for the industry to slow down in the face of safety risks, said that how the sector addresses those risks remains an open question and that the mechanism for doing so is still under discussion. His remarks suggested that leading AI companies see the framework as a starting point for debate rather than a settled model of accountability.

For analysts watching the arrangement, the accord's durability will depend less on its four pledges than on what follows them. Voluntary commitments have a poor record when commercial incentives run the other way, and the decision to place enforcement in an industry-appointed committee — with a presidential appointee still to be named — transfers the most consequential judgments to the parties with the most at stake.

The economic stakes help explain the reluctance to regulate. Trump noted that investment in data centres and related computer equipment has accelerated the broader economy and lifted equity markets, swelling investment accounts for higher-income Americans and boosting their spending. Any rulebook that slowed construction, constrained compute or limited deployment, in his view, would risk forfeiting momentum that he believes the country cannot afford to pause.

For now, the deal keeps that momentum intact while offering communities and consumers an imperfect answer to their concerns. Whether the promise of self-policing satisfies either remains the open question as the accord moves from a White House photo-op to the slower work of independent audits, board committees and the still-unnamed overseer Trump said he would select.