The high-risk classifier, on your own site.
A free, self contained three question tool that tells a reader whether an AI system is likely high-risk under Annex III of the EU AI Act. One line of HTML. No account, no API key, no cookies and no outside dependencies. Every embed carries an attribution link back to agentliability.eu.
Live, exactly as it embeds.
The classifier walks a reader through three plain questions drawn from Annex III of Regulation (EU) 2024/1689 and returns one of three editorial verdicts: likely high-risk, possibly high-risk, or likely not high-risk. Each verdict links to the relevant Article 26 or Article 27 guidance on this site.
It is meant for operators, compliance officers and advisers who want a first read on one deployment. It is an editorial tool, not legal advice.
How to embed it.
Paste this snippet wherever the classifier should appear.
<iframe src="https://agentliability.eu/widget/embed.html"
width="100%"
height="500"
frameborder="0"
title="EU AI Act High-Risk Classifier by Agent Liability EU"
style="border: 1px solid #e0e0e0; border-radius: 4px; max-width: 720px;">
</iframe>
| Width | Set to 100% for responsive containers. The widget is usable from 320 to 800 pixels wide. |
| Height | 500 pixels covers the three questions and the verdict. In a very narrow layout, 560 gives more room. |
| Dependencies | None. No web fonts, no external CSS, no external scripts. It loads from agentliability.eu directly. |
| Data sent | None. The classifier runs entirely in the browser and sends no form data anywhere. |
| File size | Under 25 KB. |
Licence and updates.
The widget is published under the Creative Commons Attribution 4.0 International licence. You may embed it on any site, commercial or not, with or without modification, as long as the attribution built into the widget footer stays intact. No separate backlink from the host page is required; a text credit on your page is welcome but not a condition of use.
The classifier follows the AI Act text and Annex III. It is updated when the Commission, the AI Office or the standardisation bodies publish guidance that materially changes the scope of the high-risk categories. Because the embed loads directly from agentliability.eu, your embed receives updates with no change to your code. If a version ever changes structurally, the earlier version stays at its original address.
If the classifier returns an unexpected result, if you have found an error in relation to Annex III, or if you have a suggestion for the question logic, write to the editors. Substantive feedback is acknowledged within five working days.