AI now sorts résumés, ranks candidates, targets job ads, and flags who moves to the next round. If your hiring process uses any of it, the EU AI Act has moved from background reading to operational reality. Here is what changed, what it means for your operating model, and what you should be doing about it.12
The EU AI Act is Regulation (EU) 2024/1689 — the first-ever comprehensive legal framework on AI worldwide, laying down harmonised rules on artificial intelligence. It sets out risk-based rules for AI developers and deployers regarding specific uses of AI, and its stated aim is to foster trustworthy AI in Europe.3
For recruiters and hiring managers, the headline is simple: most of the AI you touch in hiring sits in the highest-risk tier the law defines. This guide breaks down the classification, the deadlines, the obligations, and how the Act stacks on top of GDPR — with a specific eye on the Netherlands and the EU.12
Why hiring AI is treated as high-risk
Under the EU AI Act, AI systems used in employment decisions fall into the high-risk category. That covers recruitment, selection, targeted job advertising, candidate evaluation, performance monitoring, and certain decisions about compliance, contract terms or termination. AI used in recruitment and selection is classified as high-risk, and the same classification reaches into promotion and dismissal.124
The reasoning is about consequences. Decisions about who is hired, promoted, or fired have significant impact on the course of individuals' lives. AI systems can amplify existing biases and automate discrimination at scale — which is precisely why the EU subjected these applications to the strictest requirements it has.5
AI in recruitment and selection is one of the most regulated application areas under the AI Act.
AI Act Blog (NL)
The dates that matter
Two deadlines should be on your calendar. One has already passed. The others are coming.21
- Since 2 February 2025: the use of AI for emotion recognition during job interviews is prohibited under the AI Act.
- Starting 2 August 2026: AI hiring tools will need mandatory risk assessments, technical documentation, bias testing, human oversight, transparency disclosures, and continuous monitoring.
- Under current Commission planning, many Annex III high-risk rules apply from 2 December 2027.
The emotion-recognition ban is not a future risk — it is in force. If any tool in your interview stack claims to read a candidate's emotional state, that is already off the table.2
What compliance actually requires
From 2 August 2026, each high-risk hiring tool will need mandatory risk assessments, technical documentation, bias testing, human oversight, transparency disclosures, and continuous monitoring. In practice that means heavy obligations: technical documentation, human oversight, transparency to applicants, and more.14
Bias audits and testing
Bias testing is named explicitly among the 2026 obligations. Given that the entire premise for the high-risk classification is the danger of amplifying existing biases and automating discrimination at scale, expect bias auditing to be the part regulators look at hardest.15
Transparency to candidates
Transparency disclosures to applicants are a standing requirement, not an optional courtesy. Candidates evaluated by a high-risk system are owed transparency about it.14
Human oversight and monitoring
Human oversight and continuous monitoring sit at the core of the obligations. A high-risk hiring tool cannot run as a closed loop that decides on its own — there must be meaningful human review, and the system's behaviour must be watched over time.1
Compliance is the deployer's responsibility
This is the point most often missed. It does not matter whether you built the technology. If your business deploys it, compliance is your responsibility — even if the platform vendor says otherwise. A vendor's reassurance is not a legal shield.1
For staffing businesses, Employers of Record, and workforce platforms — which sit between employers, technology, and workers — the obligations are more demanding than they are for a single corporate HR department. The more parties you sit between, the more compliance you carry.1
How this connects to GDPR
If you have done GDPR work, this is the next layer, not a replacement for it. GDPR required you to rethink how you handle personal data. The EU AI Act requires you to rethink how you use the tools that process it. Candidate screening involves both: the data, and the system making sense of it.1
The Netherlands: regulators are already watching
Dutch supervision is active. During the first AI Supervision Congress in December 2025, organized by the Netherlands Authority for Digital Infrastructure (RDI) and the Dutch Data Protection Authority (AP), session 8 was dedicated entirely to AI in recruitment and selection. The message was clear: this is one of the most regulated application areas under the AI Act.2
The Dutch Data Protection Authority is direct about it. In its sixth report on the use of AI in the Netherlands, the AP states that AI is now widely used in recruitment and selection, that many of those applications fall under the high-risk category — and that the recruitment sector is insufficiently aware of this. The AP identifies three structural mistakes organizations make, the first being unawareness of the high-risk classification itself.4
Getting ahead of the deadline
You do not have to wait. The European Commission launched the AI Pact, a voluntary initiative inviting AI providers and deployers from Europe and beyond to comply with the key obligations of the AI Act ahead of time. For hiring teams, early compliance is also the cheapest way to find the gaps before they become findings.3
Key Takeaway
AI in recruitment and selection is high-risk under the EU AI Act, and the responsibility lands on you as the deployer — not the vendor. Emotion recognition in interviews is already banned; from 2 August 2026, your hiring tools need risk assessments, technical documentation, bias testing, human oversight, transparency to candidates, and continuous monitoring. In the Netherlands, the AP has flagged that the sector is not yet aware of any of this. The work to do now: inventory every AI tool in your hiring funnel, confirm which sit in the high-risk tier, and start closing the gaps before the date does it for you.214
- Inventory every AI system in your hiring funnel: sourcing, ad targeting, screening, ranking, interview tools.
- Remove or replace anything performing emotion recognition in interviews — it is already prohibited.
- Confirm which tools fall in the high-risk category and document the assessment.
- Demand bias testing, technical documentation, and human-oversight provisions — do not rely on vendor assurances.
- Build transparency notices for candidates and a continuous monitoring process ahead of 2 August 2026.