The heads of six leading AI developers signed a voluntary pledge at the White House Tuesday to put internal safety controls and independent audits around their most advanced models, ABC News reported. The document creates no regulator and no legally enforceable standard, and President Donald Trump said he prefers that the industry police itself.

What Happened

The document is titled "The White House Accord on Superintelligence: A Joint Commitment on Frontier SI Responsibilities." It was signed by Trump, Anthropic CEO Dario Amodei, OpenAI president Greg Brockman, Google CEO Sundar Pichai, Meta CEO Mark Zuckerberg, Nvidia CEO Jensen Huang and Elon Musk for xAI. ABC reported that Palantir's Alex Karp and Shyam Sankar, Amazon's Jeff Bezos, House Speaker Mike Johnson and Cabinet members also attended.

According to The Register, the commitments come in four layers. Companies will run internal controls that track what their models can do in areas such as cyber, biological and chemical threats; give internal teams the authority to check that those controls work and fix problems; bring in independent external auditors; and set up board committees to review the auditors' findings. The signatories also agreed to meet regularly to develop shared safety standards and best practices. The text says it may make sense to write the safeguards into law over time, without offering a timeline.

RelatedNew York City Comptroller Urges Anthropic to Delay IPO Over Safety and Governance Concerns

Johnson described the accord as a voluntary statement of principles. Trump called it "morally binding." The Associated Press, in a report republished by SecurityWeek, said Trump pledged to name someone to oversee the agreement within days after consulting the industry, and suggested a watchdog committee of roughly 10 people. ABC said the makeup and authority of that group were unclear.

Zuckerberg called the accord a start, built on principles and commitments to several layers of auditing and controls, ABC reported. Trump, according to the AP, said he would not hold back a technology he considers bigger than the industrial revolution.

Separately, Trump signed an executive order directing federal agencies to use "Super Intelligence" in place of "artificial intelligence" in official documents, Fox Business and The Register reported. The Register said the White House science and technology adviser has 60 days to draft a working definition.

Why It Matters

The accord is the administration's main answer so far to calls for oversight of frontier AI, and it puts the top U.S. labs and the leading chip supplier behind a single framework. Because the companies set the standards and pick the auditors, however, the arrangement leaves the key judgments with the firms being checked.

The Register noted that the text does not define "robust" controls, specify how often audits occur or on what timeline, or spell out any consequence for non-compliance. It also said many of the provisions mirror practices the companies had already announced or were putting in place.

The Criticism

Researchers quoted by the AP and Al Jazeera questioned the self-policing model. USC computer scientist Robin Jia said it was unclear whether the labs' incentives line up with the broader public interest. Alex Pascal of Harvard's Berkman Klein Center called for legal liability and regulation rather than loose language. Toby Walsh of the University of New South Wales asked, per Al Jazeera, why an industry of this size should be allowed to grade its own work.

What to Watch

The first test is who Trump names to oversee the accord, and whether the proposed committee gets any authority beyond reviewing voluntary reports. Also watch how the companies choose external auditors, what those audits cover and whether any findings are made public. Any move in Congress to codify parts of the accord would be the first step toward making the commitments enforceable.