What the new EU AI Act means for recruitment in the Netherlands

‍The EU AI Act Netherlands framework will place strict controls on AI systems that rank candidates, filter applications or influence hiring decisions. For business leaders, this is a governance issue as much as an HR concern. The act now places responsibility on the company, rather than the software that you use. Companies will need to know where AI is used, who is accountable for it and whether suppliers can provide evidence that their systems are lawful, accurate and subject to meaningful human oversight.

What does the EU AI Act actually say in relation to recruitment?

Essentially the EU artificial intelligence act is placing compliance responsibility on you as a company, the employer, as well as the software or technology that you use in your recruitment.

The Act classifies many AI systems used in recruitment as high-risk. This includes tools that target job adverts, filter applications, rank candidates or support decisions on promotion, dismissal and employee monitoring.

It places duties on both suppliers and employers. Suppliers must provide evidence on risk management, data quality, accuracy, security and human oversight. Employers must use the system as instructed, appoint competent people to supervise it and keep relevant records.

Employers must also inform workers before introducing certain high-risk AI systems in the workplace.

For business leaders, the message is simple: know where AI is used, who is accountable and whether the supplier can prove the system meets the rules.

The EU AI Act classifies many employment-related AI systems as high-risk because they can affect a person’s career, income and access to work.

This category can include systems used to:

• advertise vacancies to selected audiences;
• rank CVs or application forms;
• score online or game-based assessments;
• analyse recorded interviews;
• recommend which candidates should progress;
• evaluate employees for promotion or dismissal.

The classification depends on the system’s intended purpose and influence. A scheduling tool is unlikely to carry the same risk as software that predicts whether an applicant will perform well in a role.

For senior leaders, the distinction matters because high-risk systems face controls covering risk management, data quality, record keeping, accuracy, security, documentation and human oversight. These duties can affect procurement, legal risk, operational processes and board-level accountability.

When do the new AI EU rules apply?

The original AI Act timetable placed the main rules for high-risk recruitment systems on 2 August 2026. The European Commission has since moved the start date for these requirements to 2 December 2027, following delays in producing the standards businesses need for compliance. Other AI Act provisions, including certain transparency duties, began applying on 2 August 2026.

This extra time should be treated as an implementation period. Mapping systems, revising contracts and testing for unfair outcomes can take months, particularly in businesses with several offices, hiring teams or technology suppliers.

The EU AI Act Netherlands timetable also sits alongside existing privacy and employment law. The General Data Protection Regulation already restricts certain decisions made solely through automated processing when they have legal or similarly significant effects. A delayed AI Act deadline does not suspend those duties.

Who regulates AI recruitment in the Netherlands?

The Netherlands is developing its national enforcement structure through the Dutch AI Act Implementation Act.

Under the draft published in April 2026, the Dutch Data Protection Authority, known as the Autoriteit Persoonsgegevens or AP, is proposed as a central supervisory authority. Other sector regulators may also have responsibilities. As of 6 August 2026, this national legislation is still part of the implementation process, so companies should monitor the final allocation of powers.

The AP already provides guidance on automated recruitment assessments. It warns that online and game-based tests can involve automated decision-making and the processing of personal data. Businesses must understand what an assessment measures, how the result is produced and how it affects the candidate.

The practical point for directors and senior managers is clear: purchasing a product from a recognised supplier does not transfer responsibility for how it is used.

What should business leaders review?

Start with a company-wide inventory of every tool used between job advertising and the final hiring decision. Include features inside applicant tracking systems, video platforms, assessment products and wider HR software. AI functions are sometimes added through routine software updates without receiving a separate internal review.

For each system, leaders should establish:

1. What decision it influences. Record whether the output filters, ranks, recommends or directly rejects candidates.
2. What data it uses. Check for inferred personality traits, behavioural data, biometric information and proxies for protected characteristics.
3. Who is accountable. Assign a named owner with authority to stop or change the system.
4. How performance is tested. Review accuracy and selection rates across relevant candidate groups.
5. What the supplier will provide. Contracts should cover documentation, audit access, incident reporting, system changes and data retention.

Human oversight must be genuine. Asking a recruiter to approve an AI-generated shortlist adds little protection when that person cannot understand or challenge the ranking.

Veremark's review of How to find the real players, explains how a strong CV has become almost worthless in the age of AI, and how you can best identify the strong applicants.

Does the Act ban AI recruitment tools?

No. The EU AI Act Netherlands regime allows high-risk recruitment systems when the relevant controls are met.

This creates a clear procurement test. A supplier should be able to explain the system’s intended purpose, data sources, known limitations, testing process and human oversight features. Statements that an algorithm is objective or removes bias are insufficient without supporting evidence.

Systems that infer emotions from workplace data face particularly serious restrictions under the Act. Companies considering facial, voice or behavioural analysis should obtain specialist legal and technical advice before deployment.

What should boards and senior teams do now?

Assign responsibility to a named senior owner and involve HR, legal, privacy, procurement, information security and internal audit. Give priority to tools that reject candidates, produce rankings or evaluate video and assessment data.

Boards should also ask whether the business can answer three basic questions: Where is AI used in hiring? Who can explain its decisions? What happens when the system produces an unfair or inaccurate result?

The EU AI Act Netherlands rules will make recruitment technology more accountable. They will also expose weak procurement, unclear ownership and poorly understood screening processes. Companies that act now will be better placed to manage regulatory risk and defend the quality of their hiring decisions.

Veremark's guide to outsmarting AI job applications also shows how you can protect your business from fake applicants. A sensible starting point: use technology to support verification and screening while retaining informed human judgement over the final decision.