Podcast
2026
Published
by
Benjamin Zamble
THE CONVERSATION
In July, ZAGR took part in the podcast "Rekrutteringsrådet", hosted by Sverre Haugen at MeyerHaugen. The topic was whether organizations use artificial intelligence lawfully in the recruitment process. The conversation covered where the line sits between smart use and regulatory breach, and what happens when an AI system ranks or filters candidates.
→ Click here to listen to the episode at "Rekrutteringsrådet" on Spotify
THE REGULATION THAT APPLIES TODAY
The high-risk obligations under the AI Act, which cover recruitment under Annex III, take effect in December 2027. Three other legal frameworks already apply. The Norwegian Equality and Anti-Discrimination Act has prohibited indirect discrimination since 2018, and an imprecise ranking tool can fall directly under that prohibition. The duty to investigate and report under section 26 of that Act requires employers to examine, analyze, and document discrimination risk, and recruitment is explicitly named as an area the duty covers. GDPR Article 22 gives candidates the right to demand meaningful human review when a decision has a significant effect on them, which a hiring decision almost always has.
THE THRESHOLD FOR HIGH RISK
The threshold captures more systems than many expect. It is sufficient for a system to rank, filter, or score candidates for the high-risk requirements to apply. Human oversight is how Article 14 is satisfied. That oversight must be genuine, with actual authority and competence to change the outcome of the system's assessment.
ZAGR'S ASSESSMENT
Mapping existing AI tools, classifying their risk level, and documenting testing against discrimination risk are the steps ZAGR recommends organizations start with. ISO/IEC 42001 serves as the operative framework while the harmonized EU standards are still pending.
