Trump's AI Safety Deal: Tech CEOs Sign Self-Regulation Accord

The full text of President Donald Trump's "morally binding" AI safety deal is now public, and it puts the policing of frontier artificial intelligence squarely in the hands of the executives building it.

The accord, officially titled the Joint Commitment On Frontier Responsibilities, was shared online by tech founder and presidential advisor David Sacks. It has been signed by the leaders of the most powerful AI companies in the United States: Google's Sundar Pichai, Anthropic's Dario Amodei, Meta's Mark Zuckerberg, OpenAI's Greg Brockman, xAI's Elon Musk, and Nvidia's Jensen Huang.

The signing follows a high-profile gathering at the White House. On Tuesday, September 29, 2026, Trump met with tech executives in the East Room alongside House Speaker Mike Johnson. The attendee list read like a map of AI power: Microsoft CEO Satya Nadella, OpenAI president and co-founder Greg Brockman, Nvidia CEO Jensen Huang, AMD chair and CEO Lisa Su, Palo Alto Networks CEO Nikesh Arora, and Tesla CEO Elon Musk sat with the president. A photographer for The Washington Post captured the tableau — an image some observers compared, only half-jokingly, to "The Last Supper."

The meeting came after what The Verge describes as a surge in concern about AI potentially posing a threat to humans. That concern now has a formal answer: a voluntary commitment with no statutory force, described as "morally binding" rather than legally enforceable.

What the companies signed

The core language of the commitment frames safety as a corporate duty. "In order to build a positive future for the American people and the world, we believe every company is responsible for developing" its AI technology responsibly, the accord's opening passage states.

The signatories represent the full stack of frontier AI development. Pichai's Google and Zuckerberg's Meta train some of the largest foundation models in the world. Amodei's Anthropic has built its brand around safety research. Brockman signed for OpenAI, whose models are among the most widely deployed globally. Musk's xAI is a newer entrant racing to compete. Huang's Nvidia supplies the compute that makes all of it possible — the chips on which every frontier model in the deal depends.

That last signature matters. Nvidia is not primarily a model developer, and its inclusion signals that the accord treats the entire AI supply chain — from silicon to systems — as within the scope of self-regulation.

Why the structure matters

The deal's central design choice is self-regulation. Rather than creating a new federal agency, mandating third-party audits, or codifying enforcement penalties, the Joint Commitment asks companies to police themselves. The phrase "morally binding," used in describing the agreement, does the legal work of explaining what happens if a signatory falls short: nothing automatic.

This is the trade the industry has accepted. In exchange for a seat at the table — literally, in the East Room — the companies commit to responsibilities they define and administer. For the White House, the arrangement delivers a headline response to mounting public anxiety about AI risk without legislation. For the companies, it buys regulatory breathing room in a market where frontier development is the most expensive competitive race in technology history.

David Sacks' role in publishing the accord underscores the arrangement's character. Sacks is simultaneously a tech founder and a presidential advisor — a figure who spans both the industry doing the regulating and the government declining to do it.

The context

The signing caps a period of escalating alarm about advanced AI. As The Verge notes, the White House gathering itself followed "a surge in concern about AI potentially posing a threat to humans" — language that reflects how far the mainstream debate has moved from job displacement toward existential risk.

The presence of Satya Nadella and Lisa Su at the White House event, even as their companies' chief executives occupy different positions in the AI stack, points to how broadly the administration convened the industry. Microsoft did not appear among the named signatories of the commitment as shared by Sacks; Brockman, OpenAI's president, signed instead of CEO Sam Altman.

House Speaker Mike Johnson's attendance signals congressional awareness of the deal, though the accord itself is an executive-branch arrangement with private companies rather than legislation.

What to watch

The Joint Commitment On Frontier Responsibilities now exists as a public document, posted by Sacks and signed by six of the most consequential figures in AI. The open questions are the ones the accord's structure leaves unanswered: what verification looks like, whether signatories publish evidence of compliance, and what — if anything — happens when a company's commercial incentives collide with its pledge.

With every major frontier lab and its chief chip supplier signed on, the industry has told the public it can be trusted to manage the technology's risks itself. The next test of the deal will come the first time that promise costs one of the signatories something.