---
title: "Which risk class is my AI system? AI Act check | Pitch"
description: "Answer a few questions and get a first indication of your AI system’s risk class under the EU AI Act: prohibited, high-risk, transparency or minimal, with your…"
url: https://pitch.law/checks/ai-act
---

[pitch](https://pitch.law/)[AI Act](https://pitch.law/goals/ai-act)Which risk class is my AI system?

Free check

# Which risk class is my AI system?

A short list of questions on prohibited practices (Art. 5), high-risk systems (Art. 6, Annexes I and III), transparency duties (Art. 50), general-purpose AI and your role. Nothing is stored or sent.

Pitch's free AI Act check sorts an AI system into prohibited, high-risk, limited or minimal risk, from a short list of questions on its purpose, area of use and your role. Each result names the article behind it and the dates that apply. It is a first indication, and your answers stay in your browser.

This is a first indication, not legal advice. A lawyer checks your real situation.

This check needs JavaScript to run in your browser. The rules it applies are listed below, so you can also work through them by hand.

## The rules this check applies

For a lawyer to review and for you to follow by hand. Articles are those of the AI Act, Regulation (EU) 2024/1689, as amended by Regulation (EU) 2026/1744.

1. Not placed on the market, put into service or used in the EU: probably not covered (Art. 2(1)). Military, defence and national security, scientific research and development only, and purely personal non-professional use are outside the Act (Art. 2(3), 2(6), 2(10)).
2. Any practice in Art. 5(1) (a to h), or the two practices added by Regulation (EU) 2026/1744 from 2 December 2026: prohibited. This ends the check.
3. High-risk, route 1: a safety component of, or itself, a product under the Annex I product laws that needs third-party conformity assessment (Art. 6(1)). Applies from 2 August 2028.
4. High-risk, route 2: intended for use in one of the eight Annex III areas (Art. 6(2)). If it profiles natural persons it is always high-risk. If not, it is high-risk unless one of the four conditions of Art. 6(3) applies; then it may be not high-risk, but the provider must document that assessment before placing it on the market and register the system (Art. 6(4), 49(2)): it depends. Applies from 2 December 2027.
5. Not prohibited and not high-risk, with a duty under Art. 50 (chatbots, synthetic output, emotion recognition or biometric categorisation, deepfakes): limited risk, with transparency duties from 2 August 2026.
6. None of those: minimal risk, with no specific duties besides AI literacy (Art. 4, since 2 February 2025).
7. On top of any class: a provider of a general-purpose AI model has the duties of Art. 53 (since 2 August 2025), and of Art. 55 when the model has systemic risk (Art. 51(2): training compute above 10^25 floating-point operations, or designated).
8. Role (Art. 3): provider, deployer, importer or distributor. A deployer that puts its name on a high-risk system, substantially modifies it, or changes its purpose so that it becomes high-risk is treated as the provider (Art. 25(1)).

## Frequently asked questions

### Is this check legal advice?

No. It is a first indication from a short list of rules, each with the article it comes from. It cannot see your real situation, such as a group structure, the exact wording of your activity or documents. A lawyer checks that before you rely on the result.

### Does the check store or send my answers?

No. Your answers stay in your browser: no cookies, nothing stored, nothing sent. Only if you press the ask button at the end does the contact form open with a one-line summary, and nothing reaches us until you send that form yourself.

### When do the high-risk rules of the AI Act apply?

From 2 December 2027 for stand-alone systems in the Annex III areas and from 2 August 2028 for AI in products under the Annex I laws, after Regulation (EU) 2026/1744 moved the dates. The prohibitions and the transparency duties were not postponed and already apply.

### Why can the same AI model get a different risk class?

Because the class follows the intended use, not the technology. The same model can be minimal risk in a spam filter and high-risk when it screens job applicants. That is why the check asks what the system is for, in which area, and whether it profiles people.

### Do I have to do anything if my system is minimal risk?

Yes, one thing applies to every provider and deployer: AI literacy under Article 4. You take measures so that the people working with AI for you understand it well enough for their task. It is wise to also keep a short classification note for each system.

## Official sources

- [Regulation (EU) 2024/1689, the AI Act (EUR-Lex) ↗](https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32024R1689)
- [Regulation (EU) 2026/1744, the Digital Omnibus on AI (EUR-Lex) ↗](https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32026R1744)
- [AI Act: the European Commission’s regulatory framework page ↗](https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai)

## Go deeper

- [Our AI Act compliance service →](https://pitch.law/solutions/ai-act-compliance)
- [Goal: the AI Act →](https://pitch.law/goals/ai-act)
- [Guide: AI Act risk classification explained →](https://pitch.law/academy/ai-act-risk-classification-explained)
- [Guide: provider or deployer under the AI Act →](https://pitch.law/academy/provider-vs-deployer-ai-act)
- [Guide: the AI Act timeline →](https://pitch.law/academy/ai-act-compliance-timeline)
- [Guide: the AI literacy obligation →](https://pitch.law/academy/ai-literacy-obligation-businesses-overlooking)

## Let’s talk

Tell us what you’re working on. We’ll get the right person on it and reply within one business day. The first chat is free.

[Schedule a meeting](https://pitch.law/schedule)
