What is a high-risk AI system? The EU AI Act classification, defined

A high-risk AI system is the EU Artificial Intelligence Act's category for AI that is a safety component of, or is itself, a product under EU harmonisation law that needs third-party conformity assessment, or that is used in one of eight sensitive areas listed in Annex III. Such systems must meet the Act's Chapter III requirements before they reach the EU market.

Two routes into the category

Article 6 of Regulation (EU) 2024/1689 sets out both. The first covers AI built into regulated products listed in Annex I, where the product must pass a third-party conformity assessment. The second covers uses listed in Annex III: biometrics; critical infrastructure; education and vocational training; employment and worker management; access to essential private and public services; law enforcement; migration, asylum and border control; and the administration of justice and democratic processes.

Annex III status is not automatic. Under Article 6(3), a listed system is not high-risk if it poses no significant risk of harm to health, safety or fundamental rights and meets one of four narrow conditions, such as performing only a narrow procedural task. One override cancels those exceptions: a system that profiles natural persons is always high-risk.

What the label requires

Chapter III, Section 2 sets the requirements: a risk management system (Article 9), data and data governance (Article 10), technical documentation (Article 11), record-keeping (Article 12), transparency to deployers (Article 13), human oversight (Article 14), and accuracy, robustness and cybersecurity (Article 15).

Article 15 is the security team's article. It requires high-risk systems to resist attempts by unauthorized third parties to alter their use, outputs or performance, and, where appropriate, to include measures against data poisoning, model poisoning, adversarial examples, confidentiality attacks and model flaws.

When it applies

As adopted, the Act set 2 August 2026 as its general application date, with Article 6(1) product rules following on 2 August 2027. The AI Omnibus amendment pushed back both high-risk deadlines. The European Commission's notice of 27 July 2026 says Annex III rules apply from 2 December 2027, and rules for high-risk AI embedded in Annex I products from 2 August 2028.

Why it matters to a CISO

Classification follows the intended purpose, so the same model can be high-risk in a hiring tool and outside scope in a meeting summarizer. The extra time before the dates bite is the window to inventory systems against Annex III, map Article 15's attack list to existing threat models, and agree with legal who decides on Article 6(3) exceptions.

Where it appears on cisonews.ai

The Annex III lookup walks through all eight areas and the exceptions. What the EU AI Act requires of a CISO covers Article 15 and incident reporting, and the regulation tracker records the AI Omnibus entering into force.

Related terms

General-purpose AI model (GPAI), data poisoning and AI red teaming.