At a glance

  • The White House Accord on Super Intelligence is a one-page voluntary commitment, subtitled “Joint Commitment on Frontier Responsibilities,” that President Trump and the leaders of six companies signed at the White House on September 29, 2026.
  • The signature page carries Google, Anthropic, Meta, OpenAI, xAI and Nvidia. It asks each company for four layers of controls and audits: internal controls, an internal team to check them, an independent external auditor or evaluator, and an independent board committee.
  • It is not a law, a regulation or an executive order. It sets no penalty, no deadline, no regulator and no duty to publish results. As of November 15, 2026, we found it neither on whitehouse.gov nor in the Federal Register. The President posted both pages on Truth Social, and the White House posted them on its official X account.
On this page
  1. What is the White House Accord on Super Intelligence?
  2. Who signed the Accord?
  3. What does the Accord commit the companies to?
  4. Is the Accord legally binding?
  5. How do the four layers compare with what the law already requires?
  6. How does the Accord relate to Executive Order 14434 and the Super Intelligence Force?
  7. What has happened since it was signed?
  8. Is the Accord a license?

What is the White House Accord on Super Intelligence?

It is a voluntary pledge by six technology companies, also signed by the President, to run internal and external checks on their most capable AI models. It was signed on September 29, 2026, the same day the President signed Executive Order 14434, and it is a separate document. The order directs federal agencies. The Accord binds no one in law.

The document is short: a title, a subtitle, two paragraphs of principle, four numbered commitments and three short closing paragraphs, followed by a page of signatures. The President posted both pages on Truth Social that day, and the White House Office of Communication distributed the document as a PDF, according to the American Presidency Project. The Project, based at the University of California, Santa Barbara, archives presidential documents and holds a transcription of this one. The White House’s official X account posted both pages on September 30, 2026.

Two features of the wording shape everything after it. “Super Intelligence,” the administration’s new term, appears only in the title; the body speaks of “frontier models” and never defines them, so the text does not say which models are covered. And the commitments are framed as belief. The four layers are introduced with “we believe each company should implement,” and the closing line says each company is “committed to doing this.”

Who signed the Accord?

Seven people signed: the President and one executive from each of six companies. The names below are as printed on the signature page. The commitments are addressed to the companies, and OpenAI’s president signed for it, not its chief executive.

Signature page of the White House Accord on Super Intelligence, September 29, 2026
SignerAs printed
Donald J. TrumpPresident of the Unites States [sic]
Sundar PichaiGoogle
Dario AmodeiAnthropic
Mark ZuckerbergMeta
Greg BrockmanOpenAI
Elon MuskXAI
Jensen HuangNvidia

The list is narrower than the meeting. Press accounts place executives from Microsoft, Amazon and Palantir at the White House that day, and none of those companies is on the signature page. As of November 15, 2026, we found no report of any company signing after September 29.

That gap decides who is covered. A commitment made by six companies says nothing about a seventh, and neither Microsoft nor Amazon has made this pledge.

What does the Accord commit the companies to?

Four layers of controls and audits, “in addition to any other precautions.” In the document’s own terms, each company should:

  1. Run internal controls. “Implement robust internal controls to monitor the capabilities and alignment of its models during training and deployment around areas like cybersecurity, biosecurity, and chemical threats, and to ensure that its models do not hack or access technical systems in unintended ways.”
  2. Check the controls internally. Empower an internal team to ensure the controls, monitoring and detection are operating as intended and that issues are remediated.
  3. Bring in an outsider. “Partner with an independent external auditor or evaluator to carry out independent assessments of whether the controls, monitoring, and detection are operating as intended.”
  4. Report to the board. Designate an independent committee of the board of directors to oversee and receive reports from those teams and from the internal and external auditors and evaluators, and to ensure identified issues are remediated.

Two further sentences carry weight. “The participating companies will meet regularly to establish standards and best practices to improve the safety of their systems.” And: “Over time, it may make sense to codify these steps into laws or regulations.”

The text stops there. It does not say what makes a control “robust,” who may serve as an external auditor or how independence is tested, how often an assessment happens, or when any of it must be in place. The company is the party that partners with the auditor, and the text gives no one else a say in that choice. Reports go to the company’s own board committee. Nothing goes to the government, and nothing has to be published.

Is the Accord legally binding?

No. It is not a statute, a regulation or an executive order, and it is not written as a contract with the government. It creates no penalty for a company that does not follow it and names no agency to check. The President described it to reporters as “morally binding” and spoke of self-policing, according to Al Jazeera and CBS News.

The document says as much about itself. Its last paragraph treats legal force as something that might come later, and makes the commitment “regardless of whether this is required of companies.” A promise that anticipates being codified is, by its own account, not yet a rule.

Three practical consequences follow for anyone relying on it:

  • No one can enforce it under the Accord itself. A customer, a competitor or a member of the public has no claim under the document. Whether a company’s public safety promises could matter under existing consumer-protection or securities law is a separate question that the Accord does not address, and this guide does not answer.
  • There is no way to verify compliance from outside. The Accord does not require a company to name its auditor, publish an assessment or disclose its board committee.
  • There is no exit procedure. The text has no term, no withdrawal clause and no register of participants, so a company that stops following it has nothing to file and no one to notify.

How do the four layers compare with what the law already requires?

Parts of the Accord overlap with duties already enacted in California, New York, Illinois and the European Union, and with one bill pending in Congress. The New York and Illinois laws take effect January 1, 2027. The difference is that the laws name a regulator, set deadlines and carry penalties.

The Accord’s four layers set beside binding rules, one pending bill and the EU’s voluntary Code of Practice, as of November 15, 2026
Accord layerClosest ruleWhat the rule adds
1. Internal controls on capabilities and alignmentCalifornia SB 53 and New York’s RAISE Act: large frontier developers must publish and follow a frontier AI framework. EU AI Act Article 55: model evaluation, risk mitigation and cybersecurity for systemic-risk modelsA published document, a revenue or compute trigger, and penalties for not following it
2. Internal team to check the controlsThe California and New York frameworks must describe internal governance practicesDisclosure of the practice; no law in this comparison prescribes the team
3. Independent external auditor or evaluatorIllinois SB 315: an independent third-party auditor every year, from January 1, 2028. Pending: the FRONTIER Act (H.R. 9925) would require an annual third-party auditA mandate with a start date. California and New York require only that the framework describe third-party assessment; neither requires an audit
4. Independent board committeeNone in law. Closest: the EU’s voluntary General-Purpose AI Code of Practice, whose Measure 8.1 is presumed met when oversight of systemic-risk management is assigned to a committee of the management body in its supervisory function, such as a risk or audit committee, or to another suitable independent bodyWe found no board-committee requirement in the California, New York, Illinois or EU AI Act texts, or in the FRONTIER Act as introduced
Incident reporting to governmentCalifornia: 15 days. New York and Illinois: 72 hours. All three also require disclosure within 24 hours to an appropriate authority, such as a law enforcement agency, when an incident poses an imminent risk of death or serious physical injury. EU: without undue delay; the Code of Practice sets 2 to 15 days, by type of incident, for providers that sign its Safety and Security chapterThe Accord has no reporting duty at all

Read against that table, the Accord asks for more than any federal rule on frontier AI that we found, and less than state law on the points that make a rule checkable. Its fourth layer has no counterpart in any law in this comparison; the nearest is the board-level oversight in the EU’s voluntary Code of Practice. The state and EU rules are set out in California SB 53, New York RAISE Act requirements, EU AI Act GPAI obligations and the comparison in frontier AI laws.

How does the Accord relate to Executive Order 14434 and the Super Intelligence Force?

The order was signed the same day as the Accord and the Force was announced five days later. Neither gives the Accord legal force.

  • Executive Order 14434, signed the same day, directs federal agencies to write “Super Intelligence” in place of “artificial intelligence” in their own documents. Its text does not mention the Accord, and it places no duty on companies. See EO 14434 explained.
  • The Super Intelligence Force was announced by the President on Truth Social on October 4, 2026, in a post that cited the “Historic White House Accord on Super Intelligence,” according to NBC News and CBS News. In the passage NBC News quotes, the post says the Force will coordinate the federal government’s engagement with consumers, public interest groups, religious organizations, critical infrastructure providers and super intelligence companies. NBC News reported that the Director of National Intelligence was named to lead it. The quoted passages give it no role in checking the four layers.
  • Executive Order 14409, from June 2026, calls for a separate voluntary framework for pre-release government access to certain frontier models. It is a different instrument. See the AI executive orders list.

This matters for anyone who needs to cite the Accord. An executive order is published in the Federal Register and can be cited. As of November 15, 2026, a Federal Register search for the Accord’s title returned nothing, and we found no page for it on whitehouse.gov. The copies this guide cites are the President’s Truth Social post, the White House’s post on X and the American Presidency Project transcription.

What has happened since it was signed?

Little that can be confirmed. As of November 15, 2026:

  • We found no public announcement of a meeting of the participating companies, and no published standards or best practices from them.
  • We found no list from any signatory naming its external auditor or its board committee under the Accord.
  • Congress has not codified any of the four layers. The bills that would require audits, registration or approval are compared in AI licensing bills in Congress.

The White House AI tracker on this site is auto-updated: every 6 hours it reads the whitehouse.gov presidential actions, fact sheets and news feeds, the Federal Register’s presidential documents and the si.gov sitemap, and keeps items whose title or text matches AI terms, including “super intelligence.” If the Accord is later published in one of those places, the tracker should list it, with a link to the official copy.

Is the Accord a license?

No. A license is permission a regulator can refuse or revoke. The Accord asks nothing of the government and gives the government nothing to refuse. It is self-regulation with a presidential signature, and it leaves the question this site tracks, whether anyone needs permission to build the most capable AI, exactly where it stood: no federal permission is required. The arguments for and against changing that are in Should superintelligence need a license?

Facts here are stated as of November 15, 2026. This guide is general information, not legal advice.

Last updated Nov 15, 2026 by ZSKFL Management. Time-sensitive facts on this page are stated as of that date.

Common questions

Who signed the White House Accord on Super Intelligence?

President Trump and one executive from each of six companies: Sundar Pichai for Google, Dario Amodei for Anthropic, Mark Zuckerberg for Meta, Greg Brockman for OpenAI, Elon Musk for xAI and Jensen Huang for Nvidia. It was signed at the White House on September 29, 2026.

Is the White House Accord on Super Intelligence legally binding?

No. It is a voluntary commitment, not a statute, regulation or executive order. It sets no penalty, no deadline and no regulator, and it does not require companies to publish audit results. Its own text says it may make sense over time to codify the steps into laws or regulations. As of November 15, 2026, we found it neither on whitehouse.gov nor in the Federal Register. The President posted both pages on Truth Social, and the White House posted them on its official X account.

Did Microsoft or Amazon sign the Accord?

No. Neither company is on the signature page. Press accounts place Microsoft and Amazon executives at the White House meeting on September 29, 2026, but the six company signers are Google, Anthropic, Meta, OpenAI, xAI and Nvidia. As of November 15, 2026, we found no report of any company signing later.

Sources

  1. White House Accord on Super Intelligence, September 29, 2026, transcription by Gerhard Peters and John T. Woolley, The American Presidency Project, University of California, Santa Barbara.
  2. Post by Donald J. Trump with the Accord’s text page and signature page, Truth Social, September 29, 2026.
  3. Post by The White House (@WhiteHouse) with the Accord’s text page and signature page, X, September 30, 2026.
  4. White House unveils ‘super intelligence’ executive order and industry accord, Nextgov/FCW, September 29, 2026.
  5. How does Trump’s White House AI accord work?, Al Jazeera, September 30, 2026.
  6. White House releases ‘Accord’ between billionaire AI execs: Here’s what it says, Forbes Australia, September 30, 2026.
  7. Trump rebrands AI as ‘super intelligence’ as tech giants sign White House accord, Mashable via Yahoo News, updated September 30, 2026.
  8. Trump announces members of ‘Super Intelligence Force’ to coordinate AI policy, NBC News, October 4, 2026.
  9. Trump announces formation of AI “Super Intelligence Force”, CBS News, October 4, 2026.
  10. Executive Order 14434, Inaugurating the Era of Super Intelligence, 91 FR 63129, September 29, 2026 (Federal Register).
  11. H.R. 9925, FRONTIER Act (introduced), U.S. Government Publishing Office, July 23, 2026.
  12. Code of Practice for General-Purpose AI Models: Safety and Security Chapter, European Commission, published July 10, 2025: Measure 8.1 (responsibility allocation) and Measure 9.3 (reporting timelines).
  13. Public Act 104-0538 (SB 315), Artificial Intelligence Safety Measures Act, Illinois General Assembly, approved July 6, 2026.