AI-Powered Hiring Tools: What HR Teams Must Know Before the EU AI Act Deadline

Every major applicant tracking system vendor is pitching AI-driven resume screening, interview scoring, or candidate ranking right now. For HR teams in Europe, that vendor message now comes with a compliance question that most vendors haven't fully answered yet: does your tool qualify as high-risk AI under the EU AI Act?

The answer, in most cases, is yes.

Why Hiring AI Falls Squarely in the High-Risk Category

Annex III of the EU AI Act lists the categories of AI systems that are automatically classified as high-risk. Number four on that list is explicit: AI systems used for recruitment and selection of natural persons, including for advertising vacancies, screening or filtering applications, and evaluating candidates in the course of interviews or tests. This is not a gray area. Recruitment AI is high-risk AI.

The implications are significant. High-risk AI systems are subject to requirements that go well beyond a standard product purchase. Under Article 9, operators and deployers must implement a risk management system throughout the system's lifecycle. Article 10 requires training data to meet quality criteria and be free of discriminatory bias. Article 13 mandates transparency and logging. Article 14 requires human oversight to be built into how the system is used.

For HR teams, this is not just a procurement question. It is an operational and legal responsibility that sits squarely on your desk.

What Deployer Actually Means for HR Departments

One of the more important distinctions in the EU AI Act is between providers (those who build and sell the AI system) and deployers (those who use it in a professional context). Most HR teams using off-the-shelf recruitment AI are deployers.

Deployers are not off the hook under the EU AI Act. Article 26 sets out specific obligations for deployers of high-risk AI. You must ensure that staff who use the system receive adequate training. You must carry out a fundamental rights impact assessment before deploying certain systems. You must monitor the system for risks during use and report serious incidents to your national market surveillance authority.

This changes how HR professionals should approach vendor conversations. The question is no longer just what can this tool do. It is what documentation do you provide so we can fulfill our obligations as deployers.

The Bias Problem Is Not Going Away

The EU AI Act requirements on training data in Article 10 exist because the problem of biased AI hiring algorithms is documented and ongoing. Amazon famously scrapped its AI recruitment tool in 2018 after discovering it systematically penalized resumes that included the word womens. Similar issues have been documented with tools that use voice analysis, facial recognition, or language models trained on historical hiring data.