For HR: AI in hiring and managing people

AI used to recruit, screen, select, promote, dismiss, allocate tasks or monitor and evaluate workers is high-risk (Annex III, point 4). Most HR teams are deployers of tools they buy.

Check first

• Does any tool infer emotions from candidates or employees (for example "engagement" or "confidence" scores from video or voice)? That's prohibited at work. Switch it off.
• Does any tool rank, score or filter people? That's high-risk.
• Did you configure a general AI tool (such as a chatbot) to screen candidates? You may have become the provider of a high-risk system.

What you'll need (from 2 December 2027)

1. 1
   The vendor's instructions for use, and evidence the vendor has done its provider duties.
2. 2
   Trained people who review the AI's recommendations and can override them.
3. 3
   Workers and their representatives informed before the system is used.
4. 4
   Candidates and employees told when AI is used in decisions about them.
5. 5
   Logs kept for at least six months.
6. 6
   A process for people to challenge decisions, and to get an explanation where they have that right.

GDPR already applies today: automated decisions with significant effects (Art. 22) and a DPIA for this kind of processing.