Skip to content
fencia.

AI SYSTEM REVIEW

Classify each AI use case under the EU AI Act

The same model can carry different responsibilities in different settings. Fencia brings a system's purpose, role and risk information into one review, then shows which covered requirements may apply.

Review an AI systemSee how it works
Example data. Fencia suggests a context and leaves missing facts visible; a responsible person confirms it.

Live product

Role, risk and obligations per system

EU AI Act risk classification, system by system

Walk each AI system through the prohibited practices of Article 5, the high-risk uses of Annex III and the general-purpose AI rules. The result becomes the requirements and tasks you work on.

  1. Step two of the EU AI Act risk classifier: the prohibited practices of Article 5, from manipulative techniques and social scoring to emotion recognition at work and untargeted facial scraping, each with its paragraph.

    Article 5 first: is any use prohibited?

    The classifier lists the prohibited practices with their paragraph — manipulation, exploiting vulnerabilities, social scoring, biometric categorisation, emotion recognition at work or school — so a team rules each one out explicitly before Annex III.

    Every practice is cited to its paragraph of Article 5

  2. EU AI Act task checklist in Fencia: Article 9 risk management with four tasks — system established and documented, risks identified, mitigation adopted, testing in place — each with its paragraph and a status.

    High-risk obligations as a task list

    Articles 9 to 15 become tasks: risk management, data governance, technical documentation, record-keeping and the rest, each with the paragraph it comes from and its status.

    A task marked done still needs evidence before its requirement counts

Real Fencia screens from a test workspace. Names and data are fictional.

Before you classify

A suggested category is not a formal legal classification. Missing purpose, role or data information can leave applicability unknown.

Is the classification final?

No. It is a suggestion built from what you describe. A person confirms it, and only a confirmed classification counts in readiness.

Why does my role matter?

The EU AI Act gives providers, deployers, importers and distributors different obligations, so the same tool can mean different work for different companies.

What happens when a system is high risk?

Its obligations from Articles 9 to 15 appear as requirements and tasks, and Fencia can draft its Annex IV technical documentation.

Can the agent help?

Yes. Ask it to classify the systems that have no risk level: it proposes risk, role and data processed, and you decide what to apply.

Where classified systems live

AI INVENTORYAI Inventory

Keep one register of the AI systems you use or build: owner, purpose, role and risk information, plus candidates from connected sources to confirm.

TRACEABLE FINDINGSEvidence & Findings

Every finding comes with its source passage, requirement, severity and next action. Accept it, or reject it with a reason, and keep the decision on record.

Classify your first AI system

Describe what it does and who uses it, and see the obligations that may follow.