High-Risk AI

What is High-Risk AI?

High-Risk AI refers to AI systems that, under the AI Act, pose significant risks to safety or fundamental rights and are therefore subject to strict obligations, including risk management, technical documentation, human oversight, and conformity assessment before being placed on the market or put into service.

How the classification is determined

The AI Act defines High-Risk AI primarily through the product categories listed in Annex I and the application areas listed in Annex III, including critical infrastructure, education, employment, essential private and public services, law enforcement, migration, and the administration of justice. The classification depends on the intended use of the AI system, not on the underlying technology. High risk is not a property of the AI itself, but of its use.

Why High-Risk AI is subject to additional obligations

The decisions made by these systems can have significant effects on safety or fundamental rights—for example when screening job applications or determining access to education. For this reason, the AI Act requires additional documentation and controls before and while such systems are deployed.

Practical significance

Both Providers, who develop a High-Risk AI system, and Deployers, who use it, have specific obligations under the AI Act. The scope of these obligations differs depending on their role. These obligations apply in phases, and the implementation timeline may change through future legislative amendments without changing when an AI system is fundamentally classified as High-Risk AI. Current deadlines are maintained in the AI Act entry.

Example

An AI system that screens job applications or makes decisions about access to education may, under certain conditions, be classified as High-Risk AI.

Further reading

Related terms

→ AI Act
→ Provider
→ Deployer
→ Digital Omnibus