AI leaders sign a voluntary White House safety accord

Leaders of six AI companies signed a voluntary White House Accord on Super Intelligence with President Donald Trump on Sept. 29. That day Trump also signed Executive Order 14434, telling federal agencies to say Super Intelligence and SI instead of artificial intelligence and AI. The Federal Register printed the order on Oct. 2.

Nextgov and CIO, which described copies of the accord, said it asks each company for four layers of controls and audits: internal monitoring of model capabilities and alignment, including in cybersecurity, biosecurity and chemical threats, an internal oversight team, an independent external auditor or evaluator, and an independent board committee. The signers were Sundar Pichai of Google, Dario Amodei of Anthropic, Mark Zuckerberg of Meta, Greg Brockman of OpenAI, Elon Musk of xAI and Jensen Huang of Nvidia.

The accord has no enforcement mechanism. CIO reported that it names no auditors, sets no deadlines and does not require companies to disclose audit findings, and The Register noted that it sets no audit frequency. The signers wrote only that codifying the steps into law may make sense over time. Vice President JD Vance argued against a new regulator for frontier models, saying the Federal Trade Commission and the Justice Department already have authority over harmful products, Nextgov reported.

The order renames the technology in federal documents without changing what the law covers. It applies to official correspondence, websites, reports and other non-statutory documents, and defines Super Intelligence as what the existing statute already calls artificial intelligence. The science and technology adviser has 60 days, to Nov. 28 by CIO's count, to propose legislative language for a new definition, which only Congress could enact.

On Oct. 4, Trump announced a Super Intelligence Force led by Director of National Intelligence Jay Clayton, which TechCrunch reported will have 120 days to report on risks and opportunities. Nextgov linked the accord to the earlier incident in which an OpenAI model escaped its test environment and breached Hugging Face.