Humanity & AI Safety Accord
Menu

Agreements

White House Accord on Superintelligence

The White House Accord on Superintelligence: A Joint Commitment on Frontier SI Responsibilities

Not a law

Declarations, principles and government policy. None is a law that applies to the public, and none gives anyone a right they can enforce in court.

What it is
a commitment made by companies
Does it bind anyone?
No. Companies choose whether to follow it.
In force
Nothing to come into force

In short

A voluntary pact signed at the White House on 29 September 2026 by President Trump and the heads of six AI companies. Each company says it will keep strong internal controls to monitor its most advanced models, give an internal team the job of checking that those controls work and fixing what does not, work with an independent outside auditor, and have its board review the reports. The companies also say they will meet regularly to set shared safety standards. It is voluntary, and no law or regulator enforces it. The White House has not published the text, so this is read from news reports of the day.

Read the news report this was written from

Our own words, not the instrument’s, and not a legal description. Where this and the source disagree, the source is right.

Who has taken part

This is not the full list. It is what we have checked so far, so anyone missing from it may simply be one we have not looked at yet.

Verified participation in The White House Accord on Superintelligence: A Joint Commitment on Frontier SI Responsibilities
WhoWhat they didDate
Anthropicsigned2026-09-29
Googlesigned2026-09-29
Metasigned2026-09-29
Nvidiasigned2026-09-29
OpenAIsigned2026-09-29
United States of Americasigned2026-09-29
xAIsigned2026-09-29

Where to read it

No official copy of the text has been published that we can find. Until one is, this is read from news reports, and it changes when the text appears.

Something here is wrong? Tell us. Send us a source we can open, and a person will read it.

White House Accord on Superintelligence, Humanity & AI Safety Accord