Compliance

EU AI Act: what actually changes for recruiters

Updated: July 2026 — Annex III timeline and AI Literacy

Since August 2024, the EU AI Act has been in force. Specific obligations for recruitment AI systems (Annex III) have been postponed to 2 December 2027. By contrast, AI Literacy obligations have applied since February 2025. For recruiters, this is not a threat: it is a framework that separates serious solutions from opaque gadgets.

This article explains what actually changes for HR, Talent Acquisition, and agencies: updated timeline, obligations already in force, immediate actions, and a roadmap to stay compliant without sacrificing efficiency.

HR recruiters reviewing AI Act compliance in a modern European office.
The AI Act does not block hiring innovation: it sets transparency and human oversight standards.

The essentials in 30 seconds

  • Systems that filter, rank, or assess candidates are high-risk (Annex III, employment).
  • Annex III obligations for recruitment AI are postponed to 2 December 2027 — documenting, supervising, and tracing remains best practice now.
  • Since February 2025, AI Literacy applies: train teams that use or oversee these systems.
  • GDPR remains the foundation — the AI Act adds to it, not replaces it.
  • A transparent tool with explicit job criteria and human validation meets both frameworks.

AI Act timeline: what actually applies

The AI Act rolls out in phases: some duties are already in force; others for recruitment systems (Annex III) have been postponed.

  • February 2025: ban on unacceptable-risk AI practices and AI Literacy obligations (training and awareness for relevant staff).
  • August 2025: duties for general-purpose AI models and governance bodies.
  • 2 December 2027: specific obligations for recruitment AI systems (Annex III) — originally planned earlier, now postponed to this date.
AI Act compliance roadmap and milestones for HR teams through 2027.
AI Literacy since 2025; Annex III recruitment obligations postponed to 2 December 2027.

Which hiring tools are in scope?

Annex III explicitly covers systems used to recruit, filter, or evaluate candidates for employment access. Typically in scope:

  • CV or profile scoring engines;
  • automated interview analysis tools (text or voice);
  • pre-screening systems that rank or reject applicants;
  • platforms that prioritise talent pools algorithmically.

Six concrete obligations for recruiters

1. Transparency towards candidates

Inform them that AI is involved, at which stage, and with what impact on the decision.

2. Effective human oversight

A recruiter or hiring manager must be able to validate, override, or contest any AI recommendation.

3. Documentation and traceability

Keep evaluation criteria, model versions, and decision logs. You must explain why a candidate was ranked a certain way.

4. Data quality and representativeness

Training data must be relevant and as representative as possible to limit indirect discrimination.

5. Robustness and cybersecurity

The system must resist errors, model drift, and manipulation attempts.

6. Registration and vendor compliance

High-risk system providers must register in the EU database and supply technical documentation. Verify now that your vendor is preparing for the 2 December 2027 deadline.

HR roadmap: AI Literacy now, Annex III by 2027

  • Meet AI Literacy now (in force since February 2025): train recruiters and hiring managers who use or oversee AI systems — see how to train your team on AI recruitment.
  • Inventory all AI tools used in your hiring funnel.
  • Request AI Act compliance documentation from each vendor — do not wait until December 2027.
  • Update legal notices and the candidate journey (see our GDPR and AI in recruitment guide).
  • Formalise human oversight: who validates, within what timeframe, with what traceability.
  • Run a bias test on a diverse sample of profiles.
  • Train teams to read AI outputs as assistance, not verdicts (AI Literacy requirement).

AI Act + GDPR: two frameworks, one trust requirement

The AI Act does not replace GDPR. For a broader view on legality in Europe, read Is AI in recruitment legal in Europe?

How HiLucy meets AI Act requirements

HiLucy structures pre-screening on recruiter-defined job criteria, with explainable outputs and final human decisions — aligned with AI Literacy today and Annex III obligations by December 2027. Compare with this article's requirements, then explore the HiLucy platform.

Frequently asked questions

Does the EU AI Act ban AI in recruitment?
No. The AI Act regulates high-risk systems (Annex III) with transparency, documentation, and human oversight. It does not ban AI pre-screening when a human makes the decision.
When do AI Act recruitment duties apply?
AI Literacy (Article 4) has applied since February 2025. Annex III recruitment-specific duties are postponed to 2 December 2027 — preparation should start now.
Must both recruiters and hiring managers be trained?
Yes. Anyone who uses or relies on an AI system (summaries, shortlists, decisions) must understand what the tool does, its limits, and who decides.
Can AI reject a candidate on its own?
No. GDPR (Art. 22) and the AI Act require meaningful human oversight. With HiLucy, Lucy structures and documents; the recruiter or manager decides.

Want to move from reading to action? See how Hi Lucy automates your voice AI interviews and your approach to interviews powered by artificial intelligence.

← Articles