Wallester Business Account
job.services
Published on November 16, 2026

EU AI Act and Recruitment: What HR-Tech Teams and Recruiters Must Prepare Before December 2027

Last fact-checked: 13 November 2026 | Fact-checked by: Jueri (Jüri) Hass

The EU AI Act recruitment rules were delayed in July 2026, but they were not removed. Software that filters CVs, ranks candidates, or targets job adverts is still classed as high-risk AI, and the compliance date has moved from August 2026 to December 2027. This guide explains which recruiting tools are covered, what vendors and employers must do, and what is already banned today.

Key takeaways

  • The AI Act lists AI used for recruitment and selection as high-risk. That includes targeting job adverts, filtering applications and evaluating candidates, as well as decisions on promotion, termination, task allocation and performance monitoring.
  • The Digital Omnibus on AI, Regulation (EU) 2026/1744, was published on 24 July 2026 and entered into force on 27 July 2026. It moved the high-risk obligations for these tools from 2 August 2026 to 2 December 2027.
  • The delay changes the timetable, not the scope. A tool that was high-risk before the Omnibus is still high-risk after it.
  • Some rules already apply. Prohibited practices and the AI literacy duty have applied since 2 February 2025, and AI that infers emotions in the workplace is among the banned uses.
  • Both sides have duties. Vendors (providers) must build compliant systems. Employers and agencies that use them (deployers) must run them properly and inform people.

EU AI Act recruitment rules: which tools are high-risk

Annex III of the AI Act lists employment and worker management as a high-risk area. The listed uses fall into two groups.

Use case Covered as high-risk? Typical examples
Targeted job advertising Yes Ad platforms or ATS features that decide who sees a vacancy
Analysing and filtering applications Yes CV parsing and screening, knockout filters, match scores
Evaluating candidates Yes Ranking, skills tests scored by AI, interview analysis
Decisions on terms, promotion or termination Yes Tools that recommend who is promoted or dismissed
Task allocation based on behaviour or personal traits Yes Workforce management that assigns shifts or work by profile
Monitoring and evaluating performance and behaviour Yes Productivity scoring of employees
Purely administrative support Depends on function Interview scheduling, transcription, drafting a job description (our interpretation)

There is a narrow exception. Under Article 6(3), an Annex III system is not high-risk if it does not pose a significant risk, for example because it only performs a narrow procedural task or prepares information for a human decision. But a system that performs profiling of natural persons is always high-risk, which means most candidate-scoring tools will not qualify. A provider that relies on the exception must document its assessment.

The timeline after the Digital Omnibus

Date What applies Relevance to recruitment
2 February 2025 Prohibited practices and AI literacy duty Already in force
27 July 2026 Digital Omnibus on AI enters into force Changes the dates below
2 August 2026 AI Act generally applicable (unchanged) General rules, not the high-risk duties
2 December 2026 Two new prohibitions and a marking deadline for existing generative systems Not recruitment-specific
2 December 2027 Annex III high-risk obligations Recruitment and employment tools
2 August 2028 High-risk AI in regulated products (Annex I) Not recruitment

According to legal commentary, the delay was because the supporting infrastructure, such as harmonised standards and notified bodies, was not ready. Nothing was repealed. As one analysis puts it, a technical delay should not be mistaken for regulatory leniency.

What is already banned: emotion inference at work

One rule that matters for video interview tools applies today. The AI Act prohibits AI systems that infer the emotions of people in the workplace and in education, except for medical or safety reasons. This has applied since 2 February 2025. Tools that claim to read candidates’ emotions, stress or “engagement” from their face or voice should be reviewed now, not in 2027. Prohibited practices sit in the highest penalty tier.

What vendors (providers) must do before December 2027

Providers of high-risk recruitment AI face the heaviest workload. In summary, they must put in place:

  1. A risk management system that runs for the life of the product.
  2. Data governance, including checks of training data for bias.
  3. Technical documentation that lets authorities assess compliance.
  4. Automatic logging of the system’s operation.
  5. Instructions for use that tell employers what the tool can and cannot do.
  6. Human oversight built into the design.
  7. Appropriate accuracy, robustness and cybersecurity.
  8. A quality management system, a conformity assessment, CE marking and registration in the EU database before the system is placed on the market.

The Omnibus also added a broader legal basis for processing special-category data to detect and correct bias, which addresses a practical problem: testing a hiring tool for discrimination can require exactly the sensitive data that data protection law restricts.

What employers and agencies (deployers) must do

Companies that use a high-risk recruiting tool have their own duties. In summary:

  • Use the system according to the provider’s instructions.
  • Assign trained people to provide human oversight.
  • Make sure the input data is relevant for the intended purpose.
  • Monitor the system and keep the logs it generates.
  • Inform workers’ representatives and affected workers before putting a high-risk system into use at the workplace.
  • Inform candidates who are subject to decisions supported by a high-risk system.

An employer that builds its own tool, or changes the purpose of an existing one, can take on the provider’s duties as well. (Check Article 25 with your legal adviser.)

Penalties

The AI Act uses a tiered fine system. Prohibited practices, such as emotion inference at work, fall in the highest tier of up to 35 million euros or 7% of worldwide annual turnover. Breaches of most other obligations, including the high-risk requirements, carry fines of up to 15 million euros or 3%. For small and medium-sized enterprises, and now for small mid-cap companies as well, the lower of the two figures applies.

Where this meets pay transparency

Recruitment teams now face two EU regimes at once. The AI Act governs the tools that target, filter and score candidates. The Pay Transparency Directive governs the pay information and wording that appear in the vacancy and the offer. A single applicant tracking system needs to support both, which is why we cover the pay side in our guide to the EU Pay Transparency Directive for recruiters and HR-tech teams.

A 10-step checklist before December 2027

  1. Inventory every AI feature in your hiring stack, including those inside your ATS, job board, and sourcing tools.
  2. Classify each one against the Annex III list and record the reasoning.
  3. Stop emotion inference in interview tools now.
  4. Ask vendors for their conformity plan, technical documentation, and logging capabilities.
  5. Update contracts so responsibilities for oversight, logs, and incident reporting are clear.
  6. Name human oversight owners and train them.
  7. Test for bias on your own candidate data and keep the results.
  8. Plan candidate and worker notices and how you will answer requests for an explanation.
  9. Set a log retention period and a process for reviewing the logs.
  10. Record AI literacy measures for the staff who use these tools.

Frequently asked questions

Is AI used in recruitment high-risk under the EU AI Act?

Yes. Annex III lists AI used to recruit or select people, including targeted job adverts, application filtering, and candidate evaluation, as high-risk.

When do the EU AI Act recruitment rules apply?

The high-risk obligations apply from 2 December 2027, after the Digital Omnibus on AI moved them from 2 August 2026. The prohibitions and the AI literacy duty already apply.

Did the Digital Omnibus remove the high-risk rules for hiring?

No. It changed the application dates. The classification of recruitment AI as high-risk was not changed.

Are CV screening tools banned?

No. They are high-risk, which means they are allowed if they meet the requirements. Emotion inference in the workplace is banned.

Does the AI Act apply to companies outside the EU?

It can. It applies to providers that place systems on the EU market and to providers and deployers outside the EU where the system’s output is used in the EU. Confirm the details for your case with a legal adviser.

What are the fines?

Up to 35 million euros or 7% of worldwide turnover for prohibited practices, and up to 15 million euros or 3% for most other breaches, with the lower figure applying to SMEs.

Conclusion

The EU AI Act recruitment rules gave the market 16 more months, but the work required has not shrunk. Vendors need documentation, logging and bias testing, and employers need oversight, notices and a clear map of which tools they use. Teams that start the inventory and vendor questions now will meet December 2027 as a routine milestone, not a scramble.

Related reading

Job.Services publishes verified data for employers, recruiters, and HR-tech teams. If this helped, share it with the person who owns hiring compliance.

Sources

Build Your Brand on job.services

Take advantage of this high-authority domain before a competitor does.