Six AI companies sign White House accord promising outside audits of their models
Nvidia, OpenAI, Anthropic, Google, Meta and xAI signed a Sept. 29 accord pledging independent external audits of their advanced AI models.
Why it matters
The accord is a voluntary pledge by six major AI companies rather than an enforceable rule, and it leaves auditor selection and disclosure with the companies themselves. Regulators weighing compulsory model evaluation would have to add those requirements separately; the accord alone gives outsiders no guaranteed access to findings.
Key facts
Nvidia, OpenAI, Anthropic, Google, Meta and xAI signed an accord on Sept. 29 on their most advanced AI models.
Reported by CoinDesk
US President Donald Trump described the accord as morally binding.
Reported by CoinDesk
Vance-Law said the accord does not specify who chooses auditors, what they can examine or whether findings are made public.
Reported by CoinDesk
Vance-Law said there is no penalty for ignoring the accord.
Reported by CoinDesk
Congress passed the Sarbanes-Oxley Act, which took the choice of auditor away from management.
Reported by CoinDesk
What happened
CoinDesk reports that Nvidia, OpenAI, Anthropic, Google, Meta and xAI signed an accord on Sept. 29 promising the White House they would let independent outsiders examine their most advanced AI models, in an agreement US President Donald Trump called morally binding. Each company commits to partnering with an independent external auditor or evaluator and to having independent board members review the auditors' reports and track remediation. Samuel Vance-Law, head of research at the Decentralization Research Center, wrote for CoinDesk that the accord does not say who selects the auditors, what they may examine or whether findings are published, and carries no penalty for non-compliance. He argues that under the current wording each company picks its own auditor and sets the scope of the examination. No other outlet covered the story in the evidence, so all claims here rest on a single CoinDesk article and have not been independently confirmed.
How the story developed
- Official statement
Six major AI companies sign White House accord on model audits
Nvidia, OpenAI, Anthropic, Google, Meta and xAI pledge to allow independent external auditors or evaluators to check their most advanced models.
- First report
First report by CoinDesk
Big Tech's AI auditors will only see what they're shown
- Update
CoinDesk publishes critique of the audit accord
Samuel Vance-Law writes that the accord leaves auditor selection, scope and disclosure with the companies and carries no penalty for non-compliance.
Affected entities
Original sources
| Publisher | Report | Role | Published |
|---|---|---|---|
| CoinDesk Crypto media | Big Tech's AI auditors will only see what they're shown | Original |