UK AI Council Responds to Prime Minister’s UN Address on Global AI Safety Standards

The UK AI Council welcomes the Prime Minister's commitment to lead international efforts on AI safety standards, and the announcement of a National Centre for Information Defence to counter AI-enabled disinformation. Transparent standards are exactly the kind of framework the sector has been calling for. A patchwork of national rules serves no one, least of all the public whose trust in these systems we all depend on.

But standards are only as good as the capacity to apply them. Right now, the gap between what we are proposing to test for and who is actually equipped to test it is significant. The UK AI Security Institute has done genuinely world-leading work, but voluntary testing regimes and reports of developers declining to submit models show the limits of standards without teeth or scale. If the UK wants to be the 'honest broker' the Prime Minister describes, it must invest as seriously in competency towards training, tooling, and headcount, as it does in the diplomacy of getting a declaration signed.

We would urge the government, ahead of its G20 Presidency, to match any new international principles with a genuine plan to enforce them, with a particular focus on sustained investment in technical talent, as well as statutory powers ready for when voluntary commitments fall short and close enough partnership with industry that standards stay rigorous without becoming unworkable.