Risk Classification under the AI Act
Definition
The EU AI risk classification depends on purpose, system, and legal facts.
The legal classification according to Regulation (EU) 2024/1689 based on the specific system or model, its intended use, the role of the actor, and the relevant legal provisions. Prohibited practices, high-risk systems, special transparency obligations, and rules for GPAI models do not form a uniform linear risk scale.
Simply explained
A specific AI application is assigned to the appropriate legal rules; there is no simple uniform risk ladder for this.
Why it matters in practice
The term supports companies in selecting technologies sensibly, anchoring them organizationally, and using them competently.
Example from business
A company checks whether a system for applicant selection falls under the high-risk rules.
Limits and pitfalls
- Using a blanket four-level pyramid, mixing system and model rules, confusing provider and operator roles, or failing to reassess changes in the intended use.
Origin & history
Regulation (EU) 2024/1689 came into force on August 1, 2024, and will be applied gradually according to its transitional provisions. The applicable legal status at the time is decisive for classification.
Use in mid-sized companies
- Possible application: A company checks whether a system for applicant selection falls under the high-risk rules.
Frequently asked questions
What does risk classification under the EU AI Regulation mean?
The legal classification according to Regulation (EU) 2024/1689 based on the specific system or model, its intended use, the role of the actor, and the relevant legal provisions. Prohibited practices, high-risk systems, special transparency obligations, and rules for GPAI models do not form a uniform linear risk scale. What mistakes should be avoided? Using a blanket four-level pyramid, mixing system and model rules, confusing provider and operator roles, or failing to reassess changes in the intended use.
Sources
Source checked on 10.10.2026. Editorial application examples are hypothetical; no evidence of real customer results.
Editorially reviewed on: