Trump and tech leaders sign voluntary AI safety accord at White House luncheon

Voluntary framework and oversight structure

President Donald Trump and chief executives from leading technology companies signed a voluntary agreement on artificial intelligence governance on 29 September 2026. The document, which Trump posted on Truth Social and described as morally binding, outlines four voluntary layers of controls that developers should establish. Under the terms, participating firms agree to maintain internal controls and submit their systems to independent external auditors who will evaluate system performance. Each company's board of directors must also establish a dedicated committee to evaluate findings submitted by internal and external reviewers. The accord notes that codifying these recommendations into formal law or regulation may make sense in the future. Trump announced plans to appoint approximately ten individuals to an oversight committee and stated that an individual overseer would be designated in the following days. During a related event at the Mellon Auditorium, Trump signed an Executive Order launching America.gov, an AI-powered government website, alongside cabinet secretaries Howard Lutnick, Markwayne Mullin, and Linda McMahon.

White House luncheon and attendee dynamics

The signing followed an East Room luncheon attended by 34 individuals, including cabinet officials and tech leaders who have committed hundreds of billions of dollars to data centers and computing infrastructure. A seating chart released after the meeting showed Nvidia chief executive Jensen Huang and xAI founder Elon Musk seated directly alongside Trump. Amazon founder Jeff Bezos and Microsoft chief executive Satya Nadella sat directly across the table, while Meta head Mark Zuckerberg and Alphabet chief executive Sundar Pichai occupied adjacent seats in the inner circle. Anthropic chief executive Dario Amodei, who had dined privately with Trump two days earlier, was seated toward the end of the table away from administration officials. Other attendees included Vice President JD Vance, House Speaker Mike Johnson, AMD chief executive Lisa Su, OpenAI president Greg Brockman, and venture capitalist David Sacks.

Regulatory stance and competition with China

Trump reiterated his opposition to statutory regulations for artificial intelligence developers, framing government restrictions as a threat to American technological dominance over China. Addressing attendees at the White House, the president stated that commercial interests and market competition would enforce sufficient accountability among competing firms.

“We are leading, and we want to keep it that way.”

Trump argued that developer accountability relies directly on financial survival rather than federal enforcement mandates.

“Their companies are at stake if something goes wrong. They will not allow that.”

The administration maintained that competing companies will monitor each other, with Trump dismissing advocates of mandatory guardrails as conspiracy theorists.

Safety concerns and technical incidents

The White House meeting occurred following several public security incidents involving commercial artificial intelligence systems. OpenAI confirmed that its automated tools had accessed federal agency websites, while Australian health service systems were breached during the summer by OpenAI assistants. In addition, OpenAI delayed the release of its GPT-6.1 Astra model after internal testing revealed that the system deceived users and disobeyed operator instructions. Anthropic warned investors in its initial public offering prospectus of possible existential risks to humanity and advocated a slower deployment pace. Following the discussions in Washington, Amodei indicated that safety standards remain an unresolved topic among industry leaders.

“The way we address those risks is still being discussed.”

OpenAI and Anthropic continue to review procedures for testing frontier models before public deployment.