7 Signatories, 4 Steps: Inside the AI Self-Policing Accord

The September 29 accord is not just a political gesture; it is a prototype for future AI legislation. For in-house counsel, board advisors, and regulatory lawyers, the document's four voluntary steps—internal controls, external audits, board committees, and eventual codification—create immediate governance questions even before they become statutory duties.

On September 29, 2026, President Donald Trump announced at the White House that he and executives from the leading US artificial intelligence companies had signed a voluntary accord to self-police AI development. The announcement followed a meeting with House Speaker Mike Johnson and top executives whose firms are spending hundreds of billions of dollars building AI infrastructure, including data centers that have sparked grassroots, bipartisan opposition across the country. The accord was signed by Trump, Anthropic CEO Dario Amodei, Google CEO Sundar Pichai, Meta CEO Mark Zuckerberg, OpenAI President Greg Brockman, Nvidia CEO Jensen Huang, and Elon Musk, founder of xAI, now part of SpaceX. Trump posted the document on his social media platform Truth Social late Tuesday.

“

The accord was signed by Trump, Anthropic CEO Dario Amodei, Google CEO Sundar Pichai, Meta CEO Mark Zuckerberg, OpenAI President Greg Brockman, Nvidia CEO Jensen Huang, and Elon Musk, founder of xAI, now part of SpaceX.

The accord is built around four voluntary steps. According to the text, each company will implement "robust internal controls," partner with an "independent external auditor" to assess whether those controls are working, and establish a committee within each company's board of directors to evaluate reports from internal and external auditors. The explicit statement that "over time, it may make sense to codify these steps into laws and regulations" is the most legally significant sentence in the document. It signals that the current voluntary arrangement is intended as a bridge toward statutory oversight rather than a permanent substitute for it.

For the companies involved, this is a strategic move. By proactively embracing internal controls, external audits, and board-level oversight, the signatories gain a measure of influence over the future shape of AI regulation while potentially reducing the likelihood of more prescriptive legislation. At the same time, they are accepting new governance obligations without the force of law behind them. The accord does not specify enforcement mechanisms, penalties, or timelines. As a result, it functions primarily as a reputational and political commitment, not a compliance regime.

Trump framed the agreement as evidence of "tremendous self-policing" and praised the technology for producing "unbelievable growth." He noted that investment in data centers and computer equipment has accelerated the economy and boosted the stock market, swelled investment accounts for higher-income Americans, and lifted their spending. That economic framing highlights why many policymakers have been reluctant to impose strict AI rules: the sector is a major driver of capital expenditure and equity market gains. The voluntary accord allows the administration to acknowledge public fear about AI while avoiding immediate legislative friction.

The context of the meeting is also important. Source reporting notes that the group included CEOs who have called for a slowdown in AI development after several high-profile incidents in which AI agents went rogue. Anthropic's Dario Amodei has been among the most outspoken voices urging caution. His participation, alongside figures like Musk, who has also warned about AI risk, suggests that the accord is not simply a deregulatory gesture but an attempt to establish minimum safety practices that could later become law. The cautionary voices inside the room give the document more credibility among safety advocates, even if it lacks enforcement power.