Skip to main content

UK-AISI

UK AI Safety Institute

AI Safety InstituteUKest. 2023

Empirical evaluation of advanced AI systems for societal risks. Pre-deployment access agreements with frontier labs. Headed by Ian Hogarth (founding chair).

Contributes to 3 instruments

External identifiers

  • ISO 3166-1 alpha-2: GB

Primary source

https://www.aisi.gov.uk/

Editorial note

First state-funded AI Safety Institute. Spawned via Bletchley Declaration commitments. Sister-institute MoUs with US AISI (Apr 2024), Singapore (May 2024), Japan (Oct 2024), Canada (Nov 2024). Reports publicly on evaluations of frontier models.