High-risk AI system (EU)

Article 6, paragraph 2 of the EU AI Act says that "AI systems referred to in Annex III shall be considered to be high-risk". Annex III lists the areas: biometrics, critical infrastructure, education and vocational training, employment and workers' management, access to essential private and public services and benefits, law enforcement, migration and border control, and administration of justice and democratic processes. Article 6, paragraph 1 covers systems that are safety components of products under the Annex I legislation, or are such products themselves.

The 2026 amendment moved the application dates. According to the amending text, Chapter III applies from 2 December 2027 for Annex III systems and from 2 August 2028 for Annex I systems.

In practice

Check the use against the Annex III areas before anything else. An agent that screens candidates, ranks employees or decides access to an essential service falls in an Annex III area. An agent that summarises a meeting usually does not, although the same agent could fall in an area if its use changes.

High-risk status brings deployer duties such as using the system according to its instructions, and assigning human oversight to competent people.

Example

[Composite] A recruiter used an agent to rank applicants for interview. The ranking use is in the employment area of Annex III, so the deployer duties applied. The team added a documented human review of each ranking and a training record for the reviewers before the system was used in a live hiring round.

Common mistakes

• Judging risk by the model rather than the use.
• Missing a change of use, such as an internal tool that starts ranking people.
• Using the 2024 dates without checking the 2026 amendment.