The White House 'super intelligence' accord, explained: what six AI leaders signed and what it doesn't do
Google, Anthropic, Meta, OpenAI, xAI and Nvidia leaders signed a one-page AI safety pledge at the White House. It promises four layers of checks on frontier models. It is voluntary, and nothing in it is enforceable.
By ScoviAI Team
Key points
- Signed September 29, 2026, by President Trump and leaders of Google, Anthropic, Meta, OpenAI, xAI and Nvidia
- Companies commit to four layers of controls: internal monitoring, an internal oversight team, external auditors and an independent committee
- It is voluntary; the text says the steps may be written into law 'over time'
- The same day, an executive order told agencies to say 'super intelligence' instead of 'artificial intelligence'
At the end of September, the heads of America's biggest AI companies sat down for lunch at the White House and signed a pledge on AI safety. The White House Accord on Super Intelligence keeps coming up in AI news since: Anthropic, for example, cited it when it gave outside evaluators employee-like access to its systems. Here's what it actually says.
Who signed it?
According to Nextgov/FCW, the accord was signed by President Donald Trump and:
| Company | Signer |
|---|---|
| CEO Sundar Pichai | |
| Anthropic | CEO Dario Amodei |
| Meta | CEO Mark Zuckerberg |
| OpenAI | President Greg Brockman |
| xAI | CEO Elon Musk |
| Nvidia | CEO Jensen Huang |
Other executives attended the meeting but, according to several reports, didn't sign.
What does the accord promise?
It's a single page. Its core idea, in the accord's own words, is that AI safety "starts with every company that is training and deploying frontier models having robust internal processes and controls."
Each participating company is meant to have four layers of controls and audits:
- Internal monitoring of its models' capabilities and alignment during training and deployment, in areas like cybersecurity.
- An internal team that oversees those controls.
- Independent external auditors.
- A designated independent committee that reviews reports from the internal team and the auditors.
The companies also agreed to meet regularly to set standards and best practices. Musk described it as "generally grading each other's homework."
Is it binding?
No. The president called it "almost like a constitution" and "morally binding," but the text itself is voluntary. It says companies "should" implement the four layers, and adds: "Over time, it may make sense to codify these steps into laws or regulations." There are no penalties for not following it.
ADVERTISEMENT
That fits the administration's approach. Vice President JD Vance argued the same day that companies should "take the risk seriously" rather than ask for a regulatory regime, and he dismissed the idea of an FDA-style agency for AI. He said the FTC and the Justice Department already have power to act against harmful products.
The 'super intelligence' order
Hours later, the president signed an executive order telling federal agencies to use the term "super intelligence" instead of "artificial intelligence" in official communications, websites, reports and policy documents. It gives the president's science adviser 60 days to propose a legal definition.
What it means in practice
Supporters see the accord as a way to get fast commitments from companies without waiting for Congress. Critics call it self-policing with no teeth. The real test is what each company does next. So far the clearest example is Anthropic's arrangement with the evaluator METR. In Europe, by contrast, rules for AI models are now enforceable law with fines: see what the EU AI Act enforcement means.
Sources
ScoviAI Team is the ScoviAI newsroom. We write every article ourselves from the original announcements and link our sources. Spotted a mistake? Tell us and we'll fix it.
ADVERTISEMENT
