What is AI compliance screening?
AI compliance screening is the structured review of an AI system to determine whether it falls within regulatory scope, which actor role applies, which risk category is relevant, and which obligations and evidence requirements must be managed. Screening is the first governance step that connects a system to its regulatory obligations inside a controlled decision process.
Screening is not a final legal approval. It creates the governance basis from which obligations, evidence, and supervised approval follow. The result must be connected to the system record, the responsible reviewer, and the legal version consulted — so that the decision path remains reconstructable.
Key points
- Screening evaluates regulatory scope, prohibited-practice relevance, risk classification, actor role, obligation areas, and evidence gaps in a single governed process.
- The screening result is structured input for review — not the compliance decision itself. Human accountability remains with the responsible reviewer and supervisor.
- An informal assessment — emails, meetings, spreadsheet entries — cannot produce a reconstructable governance record. Structured screening can.
- Screening must identify what evidence exists and what is missing, so that the right roles can act on gaps before approval.
- The screening record must preserve the legal version consulted, the system context, the classification reasoning, and the reviewer attribution.
Why it matters
Without structured screening, organizations cannot demonstrate how they determined what the EU AI Act requires of a specific AI system. The risk is not only that the classification may be wrong — it is that the process cannot be reconstructed at audit. A governed screening process creates attribution, traceability, and a documented decision basis that survives personnel changes, system updates, and regulatory evolution.
How EAB approaches this
In EAB, AI compliance screening is embedded inside the full governance chain. The AI System Registry creates the system record. Guided Technical Completion captures operational detail. AI Screening evaluates scope, prohibited-practice relevance, risk classification indicators, and actor-role context. The Obligation Matrix translates the result into concrete obligation areas. Evidence Readiness surfaces gaps. Supervisor Approval closes the governance loop with an accountable, attributed decision.