Mark Zuckerberg and top AI executives sign voluntary White House safety accord

President Donald Trump gathered the most powerful names in artificial intelligence at the White House on September 29 for a luncheon that ended with a collective signature on a voluntary AI safety accord. Mark Zuckerberg, Jensen Huang, Elon Musk, and executives from OpenAI, Google, and Microsoft all put pen to paper on a document Trump described as “morally binding.”

The accord centers on enhanced safety reviews for frontier AI models, internal and external audits, and a self-regulatory framework that lets the companies police themselves rather than face government-imposed mandates.

What the accord actually does

The framework calls for rigorous internal and external reviews of advanced AI models before and after deployment. Companies that signed on are committing to safety controls, audit mechanisms, and oversight protocols designed to catch dangerous capabilities before they reach the public.

Trump compared the document to a Constitution. The “morally binding” label means the companies are pledging to follow through, but nobody goes to court if they don’t.

Trump also floated the creation of a ten-person oversight committee that would help guide future AI regulations.

Trump proposed that the industry stop using the term “artificial intelligence” altogether and adopt “super intelligence” instead.

A pattern of voluntary pledges

This is not the first time Silicon Valley has gathered at 1600 Pennsylvania Avenue to make promises about AI safety. The September 29 accord builds on a series of voluntary commitments stretching back to 2023, spanning both the current and prior administrations.

The Biden White House secured its own round of voluntary pledges from many of the same companies in July 2023, focused on watermarking AI-generated content, sharing safety research, and testing models for risks before release.

The guest list tells a story

Musk’s presence adds its own layer of intrigue. He has been one of the loudest critics of AI safety shortcuts, having co-founded OpenAI before departing and later suing the company over what he characterized as a betrayal of its original nonprofit mission. Sitting at the same table as OpenAI leadership to sign a shared safety document is, at minimum, a diplomatic achievement.

What this means for AI governance and markets

No immediate market-moving reactions followed the announcement. By championing self-regulation, the Trump administration is effectively telling the AI industry that it won’t face the kind of heavy-handed oversight that the European Union has pursued with its AI Act.

The proposed ten-person oversight committee could change the calculus if it gains real authority. A body with the power to review frontier models and recommend enforcement actions would represent a meaningful step beyond voluntary pledges.