Up to 7% of Turnover Ref: Art. 71
Non-compliance penalties can reach €35M or 7% of global revenue.
The new regulations classify recruitment AI as "High Risk." Here is the executive guide to understanding your liability, avoiding fines, and automating compliance before the deadline.
The Red Zone
Any AI system used for screening, ranking, or shortlisting candidates sits in the High-Risk zone.
Under the EU AI Act, any software used for screening, ranking, or selecting candidates is classified as High Risk. This means mandatory governance, transparency, and human oversight are no longer optional—they are federal law.
Non-compliance penalties can reach €35M or 7% of global revenue.
You must maintain automatic logs of every decision for forensic auditing.
"Black box" AI screening is effectively banned. You must explain why a candidate was rejected.
The Law: Training data must be free of bias and errors.
The Fix: BlindStairs validates stability. We test for demographic parity before a single candidate is scored.
The Law: Automatic recording of events (logs) over the system's lifetime.
The Fix: The "Black Box" Recorder. We generate an immutable, timestamped ledger of every screening decision automatically.
The Law: Humans must be able to intervene and oversee the AI.
The Fix: Co-Pilot Mode. BlindStairs doesn't decide; it surfaces insights. The human always has the final click.
The Law: Candidates must know they are interacting with AI.
The Fix: Automated Disclosure. Our candidate portal handles legal notifications and consent forms for you.
The EU AI Act is setting the global standard, similar to GDPR. NYC 144, California's draft rules, and UK guidelines are following suit.
One standard covers you globally.
Download our 12-page whitepaper: "Navigating the EU AI Act for Talent Acquisition." Includes a vendor audit checklist.
No spam. Just practical implementation guidance for HR, legal, and procurement.
Updated for the final May 2024 text of the EU AI Act.