Donald Trump and AI Companies Sign Voluntary Audit Agreement
Donald Trump and representatives of six major technology companies signed a voluntary agreement on internal controls, external audits and board oversight in AI development.
Donald Trump’s AI agreement with major technology companies is intended to create a shared voluntary framework for safety oversight in the development and deployment of artificial intelligence. Donald Trump said in Washington on September 29, 2026, that he had signed the document together with representatives of leading AI companies.
According to the AP, the signatories include Anthropic, Google, Meta, OpenAI, Nvidia and xAI. The agreement is neither federal law nor legally enforceable regulation. Trump described it as morally binding.
AI agreement: three layers of oversight
According to available reports, the document includes three basic control elements. Companies are expected to have robust internal safety controls, their procedures are to be reviewed by an independent external auditor, and reports from both internal and external audits are to be assessed by a board committee.
This arrangement is intended to create oversight at the companies themselves as well as outside their internal teams. However, available information does not indicate what exact technical tests companies will conduct, what safety metrics they will use or how often the audits are to be repeated.
It has likewise not been confirmed who will select the external auditors, how their independence will be ensured or who will finance their work. It also remains unclear whether companies will publish audit results or at least summaries of them.
Self-regulation instead of new binding regulation
The agreement’s significance lies mainly in the approach chosen. For some of the most influential U.S. AI developers, the administration preferred voluntary commitments with audits over the immediate introduction of new federal rules with an enforcement mechanism.
This also means that the agreement itself does not establish a legal obligation or confirmed penalties for noncompliance. Without the full text of the document, it is not possible to assess in detail the scope of the individual signatories’ commitments or the mechanisms intended to monitor compliance.
Reuters reported that Trump published the agreement document and that Google, Anthropic, Meta, OpenAI, X and Nvidia were among the companies that signed it. The AP lists xAI among the signatories. However, the full text of the agreement could not be found on the White House’s official website; the known details are based on reports citing the published document.
What comes next
The next important step will be the publication of an authentic complete text of the agreement, including the signature pages. Companies may then announce whom they selected as external auditors, which safety criteria they will review and what specific role their boards’ committees will receive.
It will also be important to watch whether other developers of advanced AI models join the framework and whether Congress or federal agencies propose rules that go beyond the voluntary agreement.
Sources
- Associated Press – Confirms the signing of the agreement, the names of the signatories and the three stated layers of oversight: internal controls, independent audit and board oversight.
- Reuters (distributed via Boursorama) – Confirms that Trump published the agreement document and that representatives of Google, Anthropic, Meta, OpenAI, X and Nvidia signed it.
- SecurityWeek – Original supplied source; describes the agreement’s voluntary nature and its focus on four company actions.
Verified and updated: 09/30/2026 06:22
