EU AI Act FAQ
What does the EU AI Act require for employers using AI in recruitment?
Quick answer
AI used in recruitment, CV screening, promotion decisions, and performance assessment is classified as high-risk under the EU AI Act. Irish employers using AI in these processes must: conduct a conformity assessment before deploying the AI, maintain technical documentation, implement human oversight (a qualified person who can review and override AI recommendations), provide transparency to candidates about AI use, and register the system in the EU AI system database.
The specific obligations for employers using recruitment AI
For Irish employers using AI in recruitment, EU AI Act compliance requires addressing five specific obligations. First, conformity assessment: before using an AI system for CV screening, candidate ranking, or any stage of the selection process, the employer must verify that the system has been through the required conformity assessment. If buying from a vendor, the employer needs documentary evidence from the vendor. Second, technical documentation: the employer must have access to the technical documentation for the system and must keep records of its use. Third, human oversight: a qualified HR professional or manager must be assigned as the human overseer for the AI system — not to review every AI output, but to monitor performance, detect anomalies, and retain the authority to override AI recommendations. This person must be trained to exercise oversight effectively. Fourth, candidate transparency: individuals must be informed when AI is used in a selection process that affects them, consistent with GDPR and the Act's transparency requirements. Fifth, system registration: high-risk AI systems must be registered in the EU AI system database before deployment.
What human oversight means for recruitment AI specifically
Human oversight for recruitment AI means more than having a recruiter review AI-generated shortlists before acting on them — though that is the minimum baseline. Genuine human oversight means the overseer has sufficient understanding of the AI system's methodology to identify when it may be producing biased or erroneous rankings. It means the overseer has the authority and the confidence to override AI recommendations when they seem wrong. It means the organisation has not allowed AI to become a de facto gatekeeper from which human overseers almost always defer to the output. And it means that the oversight process is documented — that there is a record of human review, of overrides, and of the reasoning behind final hiring decisions. For Irish employers, this is also where EU AI Act obligations intersect with employment law: decisions made solely by AI that affect candidates' opportunities are legally problematic under both GDPR Article 22 and the Act's high-risk AI framework.
Acuity AI advises Irish employers on high-risk AI obligations for recruitment and HR processes. See our EU AI Act compliance services.