Let AI help you hire. Keep every screen, rank, and reject lawful, bias-audited, and replay-provable.
AI agents now source, screen, rank, schedule, assess, and reject candidates. Employment AI is high-risk under the EU AI Act, and a bias audit is already mandatory for automated hiring tools in New York City. The KYE Hiring Governance Pack™ governs every automated decision at the moment it happens — so each one carries a lawful basis, a current bias-audit artefact, a human-review checkpoint where the law requires it, and a route for the candidate to contest it.
Automated hiring decisions are high-risk, regulated, and hard to defend after the fact.
- Employment AI is high-risk. The EU AI Act lists employment, worker management, and access to self-employment in Annex III — the high-risk class that carries the heaviest obligations.
- A bias audit is already mandatory. New York City's Local Law 144 bars using an automated employment decision tool to screen a candidate unless it has passed a bias audit in the last twelve months.
- Solely-automated rejections are restricted. GDPR Article 22 limits decisions with a significant effect on a person that are made with no meaningful human involvement.
- Adverse impact is measurable. The EEOC four-fifths rule and the UK Equality Act test whether a selection step disadvantages a protected group — and the evidence must hold up.
- The usual evidence is weak. Screenshots, spreadsheets, and reconstructed notes are contestable. Regulators and claimants want a signed, verifiable record produced at the moment of the decision.
A signed Evidence Pack™ on every decision — verifiable from public keys alone.
- Lawful basis before processing. No automated screen, rank, or reject runs until a lawful basis and a declared purpose are resolved. Unresolved actions are refused.
- Current bias audit, enforced. An automated employment decision tool with a missing or stale bias audit is refused at the boundary. The bias audit is held as a signed, replayable artefact.
- Human review where the law requires it. A solely-automated decision with a significant effect routes to a human-review checkpoint through the KYE™ oversight surface.
- Adverse-impact review built in. When a selection rate for a protected group falls below the four-fifths threshold, a review opens before more automated rejections proceed.
- Every rejection is contestable. An automated rejection carries an explanation and a route for the candidate to contest it through the KYE™ rights-disputes rail.
Governance at the moment of the decision — across the whole hiring lifecycle.
Every AI-assisted hiring action flows through the same gate. The pipeline binds to canonical KYE Protocol™ envelopes — no bespoke shapes per employer.
- 1 — Action proposed. An AI agent proposes a sourcing, screening, ranking, scheduling, assessment, offer, or onboarding action over candidate data.
- 2 — Authority check. The gate verifies the lawful basis, the declared purpose, and — for an automated employment decision tool — a current bias audit. No basis, no action.
- 3 — Bias and impact tested. The selection-rate ratio is checked against the four-fifths threshold; a breach opens an adverse-impact review.
- 4 — Human review where required. A solely-automated decision with a significant effect routes to a human-review checkpoint before it becomes final.
- 5 — Evidence Pack™ sealed. The decision emits a signed Evidence Pack™ binding the lawful basis, the bias-audit reference, the impact metric, and the outcome.
- 6 — Candidate contestability. An automated rejection carries an explanation and a route to contest it, recorded on the envelope.
- 7 — Quarterly attestation. The accountable owner co-signs a quarterly attestation covering the decision register, open reviews, and bias-audit currency.
Bound to the employment-AI regulatory spine.
The Pack binds the canonical KYE™ artefact set to the regulations that govern hiring AI. Every claim resolves to a control row on the bound framework.
| Framework | Control area | Pack coverage |
|---|---|---|
| EU AI Act | Annex III employment high-risk + Art. 14 human oversight | full / partial |
| NYC Local Law 144 | AEDT bias audit + candidate notice | full / partial |
| GDPR | Art. 22 solely-automated decisions | full |
| UK Equality Act 2010 | s.19 indirect discrimination | partial |
| EEOC Uniform Guidelines | Four-fifths adverse-impact rule | full |
Full coverage means every requirement in the control area is bound to a canonical KYE™ rule, dictionary term, or Evidence Pack™ contract. Partial coverage means the bound surface satisfies the control area when paired with the employer's own attestation.
Run a pilot on synthetic data — then deploy with confidence.
The Hiring Governance Pack™ ships in three tiers: a fixed-scope bias-audit diagnostic, a full pilot on synthetic candidate data, and an annual enterprise entitlement. Pricing is value-based and disclosed under NDA to qualified applicants.