Trump and Six AI CEOs Sign White House Accord on Super Intelligence With Voluntary Four-Layer Safety Framework

President Trump and the CEOs of six leading AI companies signed a voluntary safety agreement at the White House on September 30, 2026, committing to four layers of controls over their most powerful AI systems. The accord has no legal teeth — but it is now the closest thing the United States has to a national AI safety policy.

The signing brings together the heads of Google, Anthropic, Meta, OpenAI, xAI, and Nvidia under a document the White House is calling the "[White House](/government/white-house-national-ai-policy-framework) Accord on Super Intelligence." Trump described the commitments as "morally binding," framing the voluntary framework as a matter of national responsibility rather than regulatory obligation.

What the Accord Commits To

The agreement structures AI safety around four escalating layers of controls and audits:

• Layer 1 — Internal monitoring: Companies deploy internal systems to detect dangerous use cases, with a focus on cybersecurity, biosecurity, and chemical threats.
• Layer 2 — Internal oversight teams: Dedicated safety teams within each company conduct ongoing evaluations of frontier model behavior.
• Layer 3 — Independent external auditors: Third-party organizations audit safety practices and findings, separate from company leadership.
• Layer 4 — Board-level governance: Executives and board members take direct accountability for AI safety outcomes at the highest level of the organization.

The specific scope of the accord centers on three threat domains: cybersecurity (AI-enabled attacks on infrastructure), biosecurity (AI-assisted development of biological agents), and chemical threats (AI guidance on dangerous chemical synthesis).

Who Signed — and Why It Matters

The six companies in the room — Google, Anthropic, Meta, OpenAI, xAI, and Nvidia — collectively develop or supply the hardware and models underpinning nearly every high-capability AI system deployed in the United States. Having all six at the table gives the accord broad industry coverage, even if it cannot compel any of them.

According to NPR, Trump hosted the CEOs in person at the White House, a deliberate show of alignment between the administration and frontier AI developers. The framing positions the U.S. government as a convener rather than a regulator — consistent with the administration's broader posture of avoiding legislation that could constrain domestic AI development.

No Law, No Penalty — So What Does It Do?

The accord carries no enforcement mechanism. Companies that fail to implement the four-layer framework face no fines, no regulatory action, and no mandatory reporting requirements under this agreement. What it does create is a documented public commitment from the most powerful AI companies in the world — one that can be cited in congressional hearings, investor calls, and international negotiations.

For businesses: If you're procuring AI services from any of the six signatories, this accord gives you a framework to ask vendors about directly. "Are you conducting third-party audits per your White House commitments?" is now a reasonable procurement question.

For policymakers: The accord sets a floor for what "responsible AI development" looks like in the U.S. context. Any future legislation will almost certainly reference these four layers as a starting point.

For competitors not in the room: Companies developing frontier AI outside these six are not bound by the accord and face no pressure to adopt its framework — yet.

What to Watch

Congress has been working on AI legislation for two years without passing anything binding. The accord gives the administration a visible safety narrative heading into that debate. Watch whether the four-layer framework becomes the template for a future executive order with enforcement provisions, or whether it remains a handshake agreement that gets quietly retired after the signing ceremony.