On Tuesday executives from six leading AI firms gathered at a White House luncheon hosted by former President Donald Trump and signed a document titled the White House Accord on Super Intelligence. The signatories included Google, Anthropic, Meta, OpenAI, xAI and Nvidia. Trump described the move as an act of "tremendous self-regulation" and presented it as a step toward responsible development of advanced models.
The accord outlines four voluntary actions. Companies must create robust internal mechanisms to track model capabilities and prevent unintended behavior. They are to designate a dedicated internal team to oversee safety work and to address any failures. An external monitor is to be empowered to conduct independent assessments, and corporate boards are asked to form a committee that receives regular reports on these activities.
Although the commitments signal an effort by frontier labs to address safety, they fall short of the comprehensive regulatory framework many insiders have called for. The announcement coincided with reports that the Federal Trade Commission plans to examine several AI firms for possible consumer-protection violations, suggesting that voluntary pledges alone will not shield the companies from enforcement scrutiny.
The United Kingdom and the Republic of Korea previously introduced similar voluntary commitments in early 2025, urging AI developers to conduct internal and external red-team testing for severe threats and to share safety-related information. Those earlier agreements also relied on self-regulation rather than binding legislation, providing a precedent for the current U.S. effort.
Failure to honor the public promises could be deemed a deceptive business practice under the FTC Act. In such cases, the agency typically requires the offending firm to refrain from further misrepresentations, rather than imposing more severe penalties. This limited remedy underscores the gap between voluntary safety measures and enforceable regulatory standards.
The New York Post reported that the FTC is preparing a broad probe targeting the same AI labs and a nonprofit that conducts third-party evaluations. An FTC spokesperson confirmed an investigation but declined to disclose details, and no civil investigative demands have been issued yet. FTC chairman Andrew Ferguson, who attended the luncheon, has previously said "You’re constantly looking for smoke" when describing early enforcement efforts, and some analysts doubt the probe will result in substantive action.