Donald Trump released a one-page AI accord Tuesday, signed by the leaders of Google, Anthropic, Meta, OpenAI, X and Nvidia. The document asks companies to maintain internal controls focusing on cybersecurity, biosecurity and chemical threats. It also calls for an internal verification team, an external auditor and independent board oversight.
Trump described the accord as “morally binding.” He said he would set up a roughly 10-person committee to oversee the effort.
Who Signed the Accord
OpenAI was represented by President Greg Brockman. Google’s Sundar Pichai, Meta’s Mark Zuckerberg, Anthropic’s Dario Amodei, Nvidia’s Jensen Huang and Elon Musk signed. Musk’s xAI is now part of SpaceX.
The accord carries no penalties. Critics called it thin and nonbinding.
The Document’s Demands
The accord asks companies to keep tight controls around security threats. It also calls for independent oversight through an internal verification team, an external auditor and board-level checks.
The document says these steps could eventually be codified into law. That is the stated goal. Whether any company follows through is voluntary.
Why It Matters Now
The accord comes after an OpenAI breach of an Australian government site. By several accounts, the accord would not have required OpenAI to disclose the breach any sooner than it did.
Australia’s parliament has summoned OpenAI’s Altman and Anthropic’s Amodei to testify. The hearings will give lawmakers a chance to ask questions directly.
What This Means for Companies
The accord is voluntary. There is no penalty for failing to meet its terms. Companies can choose to ignore it entirely without consequence.
That is the whole point of calling it “morally binding.” Morality is not enforceable by law. Neither is this document.
The Bottom Line
Trump’s administration rejects hard AI regulation while letting executives sign a voluntary accord that critics call thin and nonbinding. The pact’s voluntary nature leaves companies free to act without fear of punishment.
The accord is a statement of intent. It is not a rule. It is not a requirement. It is a list of things CEOs have promised to do on their own.
Whether any of them actually do those things remains to be seen. The committee Trump plans to set up will watch, but it cannot compel action.
Where the paper stands
The paper backs narrow rules against direct harm, such as forcing companies to disclose safety failures they hid, and is against broad rules that hand the market to the incumbents, and against letting big companies hide their failures. Trump’s AI accord does none of these things. It is a voluntary pledge with no penalties, and it would not have forced OpenAI to disclose the Australian breach any sooner than it did.
The danger is big tech dominance, not the technology itself. Licensing regimes and compliance costs only giants can afford are a moat, not a safeguard. A voluntary accord with no teeth locks out whoever would have challenged today’s leaders. The paper supports narrow rules against direct harm because they target actual damage rather than letting companies hide their failures.
The accord is a statement of intent from executives who already lead the industry. It is not a rule. It is not a requirement. The committee Trump plans to set up will watch, but it cannot compel action. Readers should track whether the signatories actually follow through, and whether the Australian breach is disclosed sooner than it otherwise would have been. The paper will watch closely, too.
Key Facts
- Accord signatories: Google, Anthropic, Meta, OpenAI, X, Nvidia
- Representatives: Brockman for OpenAI, Pichai for Google, Zuckerberg for Meta, Amodei for Anthropic, Huang for Nvidia, Musk for xAI
- OpenAI breach: Involved an Australian government site
- Committee size: Roughly 10 people
- Status: Morally binding, no penalties
The accord exists. The committee is coming. The question of whether anything changes is still open.
Source material: “Trump Unveils 'Morally Binding' AI Accord Signed by OpenAI, Google and Nvidia,” Decrypt.
Get the Notebook.
The day's best stories and every fresh verdict, in plain English, in your inbox by seven. One email a day, no more.