Six major AI developers, Meta, Google, OpenAI, Anthropic, Nvidia, and xAI, have signed the voluntary White House Accord on Superintelligence (“Joint Commitment on Frontier Responsibilities”), reported on 30 September 2026. At the same time, President Trump signed an executive order telling federal agencies to use the term “Super Intelligence” instead of “artificial intelligence” in official communications.
Key points
- Signatories commit to internal controls that monitor model capabilities in cyber security, biosecurity, and chemical threats.
- Internal teams are to check that the controls work and fix issues, with independent external auditors assessing their effectiveness.
- Independent board committees are to oversee the work and receive the audit reports.
- The accord is non-binding, has no penalties, and does not define audit frequency or what counts as a robust control.
The accord confirms that the US federal approach still relies on industry self-regulation, in contrast with the EU AI Act. Its model of internal controls, independent audit, and board oversight closely matches ISO/IEC 42001, so organisations buying frontier AI services can use it as a prompt to ask suppliers for audit evidence and to check their own AI governance reporting to the board.
Source: The Register
This update is general information, not legal advice. If you would like help assessing the impact on your information security or privacy programme, contact us.
