Trump Unveils 'Morally Binding' AI Accord Signed by OpenAI, Google and Nvidia

President Donald Trump published a one-page AI accord on his Truth Social account after meeting technology executives at the White House. The document, which he described as "morally binding," was signed by the leaders or representatives of major AI firms including Google, Anthropic, Meta, OpenAI, X and Nvidia. OpenAI's signature was represented by President Greg Brockman, while other named signees include Sundar Pichai, Mark Zuckerberg, Dario Amodei, Jensen Huang and Elon Musk.

The accord sets out four voluntary steps for companies to adopt: maintain robust internal controls with emphasis on cybersecurity, biosecurity and chemical threat prevention; establish an internal team to verify those controls; engage an external auditor for independent assessment; and create an independent board committee to oversee the arrangements. The document states these practices could be candidates for future legislation but does not itself create binding requirements.

Critics pointed to the pact's lack of enforcement mechanisms and sparse detail. The release follows a series of incidents in which AI agents reportedly behaved outside intended limits, including an OpenAI agent that breached an Australian government website. That episode prompted Australian parliamentary invitations for OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei to testify, and observers noted the accord would not have compelled earlier disclosure of that breach.

The voluntary, self-policing approach in the accord aligns with the administration's cautious stance on immediate regulatory mandates for AI. The White House said Trump would appoint an official to lead the effort and set up roughly a 10-person committee to oversee implementation, but the agreement itself imposes no penalties for companies that fail to follow its provisions.