The EU AI Act and AI Recruiting: What It Means for Hiring Teams
The EU AI Act classifies AI systems used for recruitment - CV screening, candidate evaluation, interview analysis - as high-risk. That does not ban them; it imposes obligations: risk management, data governance, technical documentation, logging, transparency to candidates, human oversight, and accuracy monitoring. If you hire in the EU with AI tools, both you (as deployer) and your vendor (as provider) carry duties.
The high-risk obligations, in plain language
- Transparency: candidates must know an AI system is used and understand its role in the decision.
- Human oversight: qualified humans must be able to review, override, and interpret the system's outputs - automated final decisions are the fact pattern regulators target.
- Logging & traceability: the system must keep records sufficient to reconstruct how an output was produced.
- Data governance: training and operational data must be relevant, representative, and managed for bias.
- Accuracy & robustness: providers must monitor performance and correct systematic errors.
How The Cognitive's design maps to these duties
- Candidates are always told they are interviewing with an AI interviewer - no impersonation.
- Every score is traceable: quotes, timestamps, full recordings, and transcripts reconstruct any outcome end to end.
- Human oversight is structural: The Cognitive produces evidence-ranked shortlists; your team reviews and decides.
- Consistent, role-relevant inputs: interviews evaluate job-related responses under identical rubrics - not demographics or pedigree proxies.
Frequently Asked Questions
Is AI recruiting software banned under the EU AI Act?
No. Recruitment AI is 'high-risk', which means it is permitted subject to obligations: transparency to candidates, human oversight, logging, documentation, and bias-managed data. Emotion-recognition in the workplace is among the prohibited practices - which is one reason The Cognitive scores what candidates say and demonstrate, not inferred emotional states.
What should employers using AI hiring tools in the EU do now?
Inventory your AI hiring tools, ensure candidates receive clear notice, keep humans in the decision loop with authority to override, and choose vendors whose scoring is traceable and auditable. Ask vendors for their technical documentation and logging capabilities - if they can't produce them, that is your answer. General information, not legal advice.