Se rendre au contenu

Recrutement

Lynx HR — Bias Audit

Parallel anonymized scoring of applicants — flags drift between identity-bearing and identity-redacted scoring so recruiters can review for bias.

lynx_hr_recruitment_bias_audit · v19.0.1.0.0 · Premium

What this solves

Lynx HR - Bias Audit

Computes a parallel anonymized fit score for every applicant and flags drift between the identity-bearing and identity-redacted scores - so recruiters review for unconscious bias and align with NYC Local Law 144 and EU AI Act Article 26 obligations.

Automated employment decision tools are now regulated in multiple jurisdictions: NYC Local Law 144 requires bias auditing of automated hiring tools, and EU AI Act Article 26 designates them as high-risk. This module computes a second fit score with name, photo references, university bucketization, and geographic markers stripped, then flags applicants whose score moves more than 10% between the two passes for human review.

Key Features

  • Parallel anonymized scoring - identity-bearing fields removed for the second score.

  • Drift detection - 10% threshold flag with explainable diff so the reviewer sees exactly what changed.

  • Per-applicant flag - "Score changes by 0.18 when identity removed - worth reviewing."

  • HR-manager-only visibility - drift information is restricted; recruiters see only the standard score.

  • Opt-in - not auto-installed; customers in regulated markets enable it manually.

  • Regulatory alignment - structured for NYC Local Law 144 and EU AI Act Article 26 audit evidence.

Integrates With

  • hr_recruitment - applicant model being scored.

  • lynx_hr_recruitment_score - the underlying scoring engine.

Try Lynx HR — Bias Audit on your team.

Free trial, no credit card. Talk to sales when you're ready.