Skip to main content
All policies

Policies & Support

NYC Local Law 144 — bias audit status

Honest status of the independent bias audit NYC Local Law 144 requires, for the Vettika AEDT (operated by Nexprove). No independent audit has been completed. AI-assisted draft — not yet lawyer-reviewed.

vdraftEffective

Status — read first

AI-assisted best-effort draft · not legal advice · not lawyer-reviewed.

This document was drafted with AI assistance as a best-effort starting point. No licensed attorney has reviewed, approved, or certified it, and it is not safe to rely on as a binding or compliant document. A qualified attorney in each operating jurisdiction must review and finish it before it is published or enforced. Nothing here claims an audit, certification, corporate registration, or executed contract that has actually happened.

NYC Local Law 144 (effective July 5, 2023) requires an annual independent bias audit of any automated employment decision tool used to screen NYC-located candidates, plus a published candidate notice and audit summary at least 10 business days before use. No such audit has been performed for Vettika. This status must not be replaced with a fabricated audit summary — the audit has to be run by a genuinely independent third party; it cannot be self-performed or AI-generated.

Keep the candor

This page covers the automated employment decision tool operated by Vettika, a service operated by Nexprove across vettika.com and airecruiter.nexprove.com. It states plainly that no independent audit has been completed and no demographic data is collected yet. That honesty is the right posture. Naming the operating entity does not change it — an audit is a third-party act, not a company detail. When a real audit completes, the auditor's signed summary replaces the status below within 10 business days of receipt.

Current status

RequirementStatusNext step
Independent bias audit conductedNot performedEngage an independent auditor before any NYC-located candidate is screened
Audit summary published at a stable URLNot publishedThe route exists as a stub; it needs a real report
Candidate notice ≥10 business days in advancePartialThe notice panel renders on the apply page; confirm the recruiter posts it 10 business days before screening, not only at apply-time
Qualifications and characteristics describedPartialThe panel describes role-relevant skills; the recruiter surfaces the specific rubric per role
Data-source descriptionStubNeeds a full, accurate description (see the data-source note)
Accommodation alternative pathStubNeeds a real intake process

The gap that matters: the product can be pointed at NYC candidates today

The product is live and can run against NYC-located candidates with no bias audit in place. The legal exposure under LL144 sits on the employer, but Vettika enabling a non-compliant run is a facilitation risk worth closing in the product itself. Two concrete steps:

  • A hard gate or a logged attestation. Before an AEDT run that can reach NYC-located candidates, require a logged recruiter attestation: “I have completed the bias audit LL144 requires,” or “I am not screening NYC-located candidates.” Store it. This is the same in-product gate the Terms and the EU AI Act page call for.
  • Tell customers the audit has to be theirs. A generic vendor audit cannot satisfy LL144 for a customer's specific rubric — every customer needs their own audit. The sales and onboarding flow has to say this clearly.

Fix the data-source description

The AEDT data page must reconcile two claims against reality before publishing: production data lives in Neon us-east-1 (us-east-2 is the separate QA database and must not be advertised as where candidate data lives), and the “audio not retained” claim depends on the unconfirmed LiveKit egress behavior flagged in the retention schedule and subprocessor list. A published AEDT data-source description that is factually wrong is itself an LL144 notice defect.

Sequence the demographic-data collection

The audit needs demographic data (sex, race/ethnicity, and intersectional combinations, per the EEO-1 schema) to compute selection rates and impact ratios. That collection is planned but unbuilt. Until it exists, the audit can never be produced. Sequence the build so a real audit becomes possible.

Methodology (for when the audit runs)

Per LL144 §20-871, the audit computes the selection rate for each category, the impact ratio against the most-selected category, and the sample size and date range. The methodology is sound. What it cannot be is self-performed.

The auditor must be genuinely independent

Per LL144 §20-870, the auditor must have no employment, professional, or financial relationship with the employer or the vendor in the preceding 12 months, and should furnish a conflict-of-interest declaration. This is a hard human requirement — it cannot be satisfied by Vettika, by the recruiter, or by AI.

Until an audit is published

Counsel's call, and the honest options are only two: do not screen NYC-located candidates through the tool, or run only with the LL144 notice fully populated and the recruiter clearly understanding that the LL144 enforcement exposure rests with them as the employer. There is no third option that implies compliance without an actual independent audit.

A non-AI alternative is always available — see Request a non-AI alternative. To flag a decision, email support@vettika.com with the apply link or interview id.

Questions about this document? Email hello@vettika.com.
AEDT Bias Audit Summary — Vettika