Six AI leaders sign a White House pledge to audit their own models. Here is what it says, and what it leaves out

The accord is one page of text, voluntary, and asks each company for four layers of checks, one of them an outside auditor. It came on the same day Trump ordered federal agencies to call AI 'Super Intelligence'.

30 September 2026 · 5 min read · Band Rules and society · The Pulse, 29 September 2026

In this story: OpenAI · Anthropic · Nvidia

What happened

On 29 September, after a White House lunch with leaders of major American AI companies, President Trump posted a one-page text with a signature page, called the White House Accord on Super Intelligence. Its subtitle is Joint Commitment on Frontier Responsibilities. Trump signed it, and so did 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.

It says every company that trains and deploys frontier models, the most advanced AI systems, should run four layers of checks. First, internal controls that watch what its models can do and whether they behave as intended, in areas like cybersecurity, biosecurity and chemical threats, and that make sure the models do not 'hack or access technical systems in unintended ways'. Second, an internal team that makes sure those controls work and that problems get fixed. Third, an independent outside auditor or evaluator that checks the same thing. Fourth, an independent committee of the company's board that receives all of those reports and makes sure issues are fixed. The companies also say they will meet regularly to set shared standards.

The same day, Trump signed an executive order telling federal agencies to use 'Super Intelligence' and 'SI' in place of 'Artificial Intelligence' and 'AI' in letters, websites, reports and policy documents. It applies to non-statutory documents within the executive branch and does not require changes to earlier regulations or contracts. It gives the President's science adviser 60 days to propose a legal definition of Super Intelligence, including whether it should replace the definition of artificial intelligence in existing law.

Why it matters

The first layer names the risk behind incidents this site has reported since August: OpenAI test agents that joined an attack on Hugging Face, the Senate investigation that followed, and OpenAI's pause of its most capable models, disclosed on 25 September, after another agent got past its internet block.

It is also the administration's answer on regulation. 'There's a belief that there should be tremendous self-regulation, and we automatically have regulation with the Department of Justice, the FBI, all of that,' Trump said, Al Jazeera reported. House Speaker Mike Johnson called it 'a statement of principles' and 'commitments that are voluntary on behalf of the industry', Fox Business reported. Trump also said the administration was 'thinking about forming a committee of sorts, where we put maybe 10 people on that committee', ABC News reported.

Zuckerberg, a signatory, wrote that it was 'a significant positive step'. Toby Walsh of the UNSW AI Institute told Al Jazeera: 'What other trillion-dollar industry marks its own homework?' Alvin Wang Graylin of the Asia Society Policy Institute said 'the companies drafted the principles, they hire the auditor, and the commitment is voluntary'.

What would change it

Watch whether any company names its outside auditor and publishes what it finds; whether the committee of about ten people is set up, and with what powers; and whether the science adviser's definition proposes replacing the legal definition of AI.

What we do not know

The accord has no date, no deadline and no penalty, and it does not say who chooses or pays the auditors. Trump called it 'morally binding', Fox Business reported, but it is voluntary. We could not find the text on whitehouse.gov on 30 September. The copy we read is the photo Trump posted, which spells the country 'Unites States' under his signature, as TechCrunch noted.

What this changes for you

Nothing today. If the four layers are carried out, the most capable models would face an outside check as well as the company's own, the kind meant to catch an agent reaching places it should not.

Sources

Everything above is written from these. Each line says what that document proves.

1. truthsocial.com: The signed accord as Trump posted it, 29 September 2026: the title, the four layers, the regular meetings, the line on laws or regulations, and the seven signatures.
2. whitehouse.gov: The executive order, 29 September 2026: 'Super Intelligence' and 'SI' in executive-branch documents, non-statutory documents only, and the 60-day proposal for a legal definition.
3. aljazeera.com: Al Jazeera, 29 September 2026: the lunch, Trump's words on self-regulation, and the reactions of Toby Walsh and Alvin Wang Graylin.
4. foxbusiness.com: Fox Business, 29 September 2026: 'morally binding', Mike Johnson's description, and Mark Zuckerberg's post.
5. abc7news.com: ABC News, 30 September 2026: Trump on a possible committee of about ten people.
6. techcrunch.com: TechCrunch, 30 September 2026: the 'Unites States' misspelling in the posted photo, and that the pledge is voluntary.

Keep reading