# UK AI Security Institute
**Aliases:** AISI; AI Security Institute
**Entity class:** United Kingdom government research organization
**Maturity:** Developed
## Role
The [UK AI Security Institute](https://www.aisi.gov.uk/about) is a government organization within the Department for Science, Innovation and Technology. It evaluates advanced AI systems, including through work with frontier companies and allied governments. It is the state technical-evaluation layer discussed in [[articles/war with empire/The Real No Kings Moment and Who Wrote the Super-intelligence Ban|The Real No Kings Moment]], separate from the private advocacy organization [[wiki/ControlAI|ControlAI]] and the royal/Ditchley convening.
The article reports that Anthropic did not give the Institute pre-release access to Claude Mythos 5.1 in September 2026 and attributes the account of Whitehall's reaction to the *Financial Times*. The event illustrates a limit on evaluator authority when a model producer can withdraw substrate access; it does not show that the Institute wrote either proposed superintelligence bill.
## Relationships
- **Government home:** The Institute belongs to the UK Department for Science, Innovation and Technology.
- **Analytical role:** [[wiki/First-Mover Evaluation Authority|First-Mover Evaluation Authority]] explains the advantage of recognized evaluation categories; [[wiki/Certification State|Certification State]] describes the wider permission architecture.
- **Distinct counterparts:** [[wiki/ControlAI|ControlAI]] is an advocacy organization, and the [[wiki/Artificial Superintelligence Security Bill|Sobel bill]] is a private member's bill rather than an AISI act.
- **Source and route:** [[articles/war with empire/The Real No Kings Moment and Who Wrote the Super-intelligence Ban|The Real No Kings Moment]] in [[collections/War With Empire|War With Empire]].