AI Risk Assessment ToolHow we use AI
Short answer: no AI model runs on airiskassessmenttool.com. The EU AI Act screening and the AI agent risk score are fixed rules written in advance.
What the tools do, and don't do
- They apply fixed rules to your answers: the same answers always give the same result. The rules are in the page's own scripts (/screen.js, /agent-risk.js), which anyone can read.
- They don't use a language model, a chatbot, or any AI service. Nothing on this site generates text or images for you.
- They don't make decisions. A result is a screening; you and your advisers decide what to do.
What data they see
Your answers stay in your browser. The page's security policy only lets it send data to Agent Trust Cloud, and only if you tick the consent box and ask for email or a free workspace. Nothing is sent to an AI provider.
Models and providers
None. There is no AI model or AI provider behind this site.
How to turn it off
There is nothing to turn off: the site has no AI features. The AI Governance Toolkit is Word and Excel files with no AI features either.
Why we publish this
People should know when they are dealing with AI. Article 50 of the EU AI Act sets transparency duties for AI systems that interact with people or generate content, applying from 2 August 2026. Nothing here is such a system, and we say so plainly. Official text: Regulation (EU) 2024/1689, as amended by Regulation (EU) 2026/1744 (checked 28 September 2026).
If we ever add an AI feature, this page will say what it does, what data it sees, which model and provider it uses, that a person decides, and how to switch it off, before the feature goes live.