Is an AI Recruitment Tool High-Risk under the EU AI Act?

Short answer

Often potentially high-risk — if your tool evaluates or ranks candidates and materially influences hiring decisions, it typically maps to Annex III.4 (employment). That is a hypothesis until intended purpose, role, and Art. 6(3) facts are confirmed.

Standalone Annex III obligations apply from 2 December 2027 (Digital Omnibus). Article 50 transparency can still matter earlier (2 August 2026).

What Annex III.4 covers (in plain terms)

Employment-related AI that is used for recruitment or selection — including screening and ranking applications — is the classic SaaS example founders worry about.

Signals that push toward high-risk:

• CV / resume scoring or ranking
• Automated shortlisting
• Skill / fit assessment that feeds hiring decisions
• Profiling of candidates

Signals that may support an Art. 6(3) discussion (needs counsel):

• Output is purely preparatory
• Does not materially influence the decision
• A human makes the real hiring call with meaningful discretion
• System does not profile natural persons (profiling generally blocks Art. 6(3))

Questions that change the classification

1. What is the exact intended purpose as placed on the market?
2. Are you the provider or deployer?
3. Does the score/rank materially influence who gets interviewed or hired?
4. Does the system profile natural persons?
5. Can a human meaningfully override the ranking?

What to prepare (by deadline)

Practical next step

Run a free readiness brief for your recruiting product — you will get a classification hypothesis, confidence, missing facts, and dated next steps. Then hand that file to counsel.

Also see: AI recruiting checklist and human oversight SOP template. Cluster: AI Act for recruiting software.

Not legal advice. Classification depends on product-specific facts.