Two routes in
Article 6 sets out two ways a system becomes high-risk.
Annex I: product safety. The system is a product, or a safety component of a product, covered by EU harmonisation law that already requires third-party conformity assessment: machinery, medical devices, toys, lifts, vehicles and similar. The AI Act's requirements are added to the existing product regime.
Annex III: listed uses. The system is used in one of eight areas: biometrics; critical infrastructure; education and vocational training; employment and worker management; access to essential private and public services and benefits; law enforcement; migration, asylum and border control; and the administration of justice and democratic processes. Each area lists specific uses, and only those uses are caught.
The financial services cases
Annex III point 5 is the one financial services firms meet most often. It covers AI systems used to evaluate the creditworthiness of natural persons or establish their credit score, and AI systems used for risk assessment and pricing in relation to natural persons for life and health insurance. The credit point carries an express exclusion: AI systems used to detect financial fraud are not high-risk under it.
So a retail credit decisioning model is high-risk, a fraud model is not, a motor insurance pricing model is not (only life and health are listed), and a model that scores small businesses rather than natural persons is not caught by this point. Each of those conclusions should be written down, because the classification is the organisation's own and an authority may test it.
The derogation, and the profiling rule
Article 6(3) lets an Annex III system escape high-risk status if it does not pose a significant risk of harm, for example because it performs a narrow procedural task, improves the result of a human activity already completed, detects patterns without replacing human assessment, or performs a preparatory task. A provider relying on the derogation must document its assessment and still register the system.
One rule overrides the derogation entirely: an Annex III system that performs profiling of natural persons is always high-risk. Most credit and insurance pricing models profile, which is why the derogation rarely helps in financial services. The Commission published draft guidelines on classification on 19 May 2026; they are not yet final.
What the requirements are
A high-risk system must have a risk management system across its lifecycle (Article 9), data governance including examination for bias (Article 10), technical documentation (Article 11), automatic logging (Article 12), transparency and instructions for use for deployers (Article 13), designed-in human oversight (Article 14), and appropriate accuracy, robustness and cybersecurity (Article 15). The provider must also run a quality management system (Article 17), complete a conformity assessment before the system is placed on the market, register it in the EU database, and monitor it once it is in use.
Deployers have their own duties under Article 26: use the system according to its instructions, assign human oversight to people with the competence and authority to exercise it, monitor its operation, keep its logs for at least six months, and inform workers and their representatives before using it in the workplace. Some deployers, including those using credit-scoring and life and health insurance pricing systems, must also carry out a fundamental rights impact assessment (Article 27).
When it applies
The 2026 Digital Omnibus deferred the high-risk obligations: Annex III systems from 2 December 2027 (originally 2 August 2026) and Annex I systems from 2 August 2028 (originally 2 August 2027). Systems already on the market before those dates are covered only if they are later substantially modified, except that high-risk systems intended for use by public authorities must comply by 2 August 2030. The conformity assessment, the technical documentation and the fundamental rights impact assessment are the slow parts, so the deferral is less time than it looks.
Frequently asked questions
What makes an AI system high-risk under the EU AI Act? Either being a safety component of a product covered by EU product safety law (Annex I) or being used in one of the eight areas listed in Annex III, such as employment, education, credit scoring or law enforcement.
Is credit scoring high-risk? Yes, when it evaluates the creditworthiness of natural persons. AI used to detect financial fraud is expressly excluded.
Is insurance pricing high-risk? Risk assessment and pricing for life and health insurance is. Other lines of insurance are not listed.
When do the high-risk obligations apply? From 2 December 2027 for Annex III systems and 2 August 2028 for Annex I systems, after the 2026 Digital Omnibus deferral.
Can a high-risk system be reclassified as not high-risk? An Annex III system can rely on the Article 6(3) derogation if it does not pose a significant risk, but never if it profiles natural persons.
Where it sits in the register
Missing AI impact assessment process and AI system lifecycle controls gap are the gaps the high-risk requirements are built to close, and inadequate human oversight design is the Article 14 failure. The controls are the AI impact assessment process, human-in-the-loop review and AI system monitoring and logging. Classification starts from the AI inventory.