White House Accord on Superintelligence: Google, Anthropic, Meta, OpenAI, xAI and Nvidia sign voluntary four-layer controls and audits; same-day executive order renames AI 'Super Intelligence' in federal documents

Why it matters

【Key points】

On September 29, after a White House meeting with AI leaders, President Trump signed "The White House Accord on Superintelligence: A Joint Commitment on Frontier SI Responsibilities" with Sundar Pichai (Google), Dario Amodei (Anthropic), Mark Zuckerberg (Meta), Greg Brockman (OpenAI), Elon Musk (xAI) and Jensen Huang (Nvidia). The voluntary accord sets four layers: internal controls monitoring model capabilities and alignment in training and deployment; an internal team ensuring those controls work; an independent external auditor or evaluator; and an independent board committee overseeing reports and remediation. Trump called it "morally binding."

The same day, Executive Order 14434 told agencies to say "Super Intelligence" (SI) instead of "AI" in official documents, using the existing statutory AI definition for now, with 60 days to propose a federal definition.

【Verifiable】

The order is on whitehouse.gov and in the Federal Register (Oct 2). Title, signers and structure are in Speaker Johnson's press transcript and Nextgov and The Hill reports. The quoted text does not say who picks or pays auditors, how often they audit, whether reports are public, or what happens if a signer falls short.

【Our read】

The four layers mirror corporate financial governance (internal controls, internal audit, external audit, audit committee), making frontier safety a board duty, but without the common standards, auditor-independence rules and legal liability that make that model work. Since then the White House has created its SI Force and called incident reporting "not optional," so the accord looks like a floor. Under the order, "SI" in federal documents simply means today's AI; watch the definition due in late November.