AI systems
Register an AI System
Registration records enough about a system to check its scope, classify it and work out which governance applies, without asking for the same facts again.
- In the app
- Project → AI Systems → Register AI System
- Who can use it
- The number of AI Systems depends on your plan.
Form sections
- Basic information: name, owner (a project member), system objective and intended purpose.
- Provider and model: who supplies the system and which model it uses.
- AI system characteristics: autonomy level, adaptiveness, how it produces its output, and output types.
- Deployment: lifecycle stage, where it runs and the jurisdictions it serves.
- Data and usage: the data it processes, including personal or sensitive data.
- Human oversight and review: who reviews or can override its outputs.
Systems listed in the project wizard are created with a name, purpose and a few traits. Open them and complete the remaining fields before classifying.
Why the facts matter
- Characteristics such as autonomy and how output is produced decide whether it is an AI system under the EU AI Act.
- Jurisdictions and data types influence which frameworks and privacy work apply.
- Whether you are the provider or the deployer changes your responsibilities.
- Affected people, autonomy and oversight shape the assurance and approvals needed.
After you register
Choose Register AI system. The system opens at Step 0, the Scope Check, which records whether it is In scope (AI system), Out of scope (AI Act) or Scope unclear. If you override the suggested result you must give a reason.
Last reviewed 2 October 2026
Next stepAI System classificationClassification records the system's EU AI Act risk tier and the reasoning behind it. It sets the regulatory context; the Governance plan then shows the work that follows from it.