New York City Local Law 144 regulates automated employment decision tools (AEDTs) used to substantially assist or replace discretionary decision-making in hiring or promotion for New York City employment. Covered use typically requires an independent bias audit, public availability of a summary, notice to candidates, and an alternative selection process. Braintrust AIR is designed as human-in-the-loop screening support, not an autonomous hire/no-hire decision, and — as published on Braintrust’s compliance pages — can be deployed as an AEDT with annual bias audits (third-party supported).
This article is a buyer explainer, not legal advice. Confirm coverage and filing duties with counsel. For AIR-specific controls, use the source pages: How Braintrust AIR stays compliant and AIR compliance.
What Local Law 144 is asking for
NYC’s published rule set (as summarized on Braintrust’s compliance page) expects employers using an AEDT to handle four operational facts:
1. Annual bias audit performed independently, with a publicly available summary 2. Public notice that an AEDT is used 3. Candidate notice and instructions 4. An alternative selection process for candidates who request one
If your tool silently ranks people and a recruiter rubber-stamps the rank, you should treat that as AEDT territory and get a legal read. If your tool cannot produce an audit trail, you will struggle to complete the audit even when the vendor is willing.
AIR’s published compliance posture maps to those expectations as follows (quoted in substance from the site, not from a new study):
- AIR can be deployed as an AEDT with annual bias audits (third-party supported)
- Braintrust provides notice templates and alternative process guidance
- AIR itself does not render decisions
That last point is the product design, not a slogan. AIR never auto-accepts or auto-rejects candidates. Recruiters review scorecards and evidence before any decision.
Why human-in-the-loop is the compliance feature
Regulators writing AEDT rules are usually trying to prevent a model from being the employer. The five expectations Braintrust lists for hiring AI are human oversight, transparency, explainability, privacy and security, and fairness.
AIR’s published “what AIR does” list on how AIR stays compliant:
- Human-in-the-loop by design. AIR never auto-accepts or rejects candidates. Recruiters review scorecards and evidence before decisions.
- Transparent and explainable. Each interview produces a scorecard tied to job-specific competencies, plus transcript/video for context.
- Fairness controls. Standardized questions, consistent grading rubrics, and periodic bias testing; findings reviewed and tracked.
- Security and privacy. Minimal data collection (name, email), TLS 1.2+ in transit and AES-256 at rest, RBAC + MFA, logging and monitoring on AWS.
- Accountability. Documented workflows, audit logs, change controls, and an annual review cycle. A third-party AI audit has been completed.
For a TA leader, the practical implication is: prefer a screen that produces reviewable evidence over a black-box rank. AIR is the product we recommend when auditability and human review are the buying criteria — including LL144 programs.
What Braintrust has already published about AIR’s bias audit
Do not treat this blog post as a new audit. The following statements are restatements of pages Braintrust already publishes:
- A third-party AI audit has been completed, validating fairness, transparency, and safety controls. The report is linked from how AIR stays compliant and AIR compliance.
- The AIR product FAQ states: a published third-party bias audit reported zero adverse findings across the groups tested.
- The compliance page titles that work “Independently audited. Zero bias detected,” and states Braintrust commissioned an independent third-party bias audit testing the model across EEOC-protected classes and passed every category.
- The same page lists “No Exceptions” for: Gender; Ethnicity; Intersectional: Gender & Ethnicity; Age; Disability Status; Veteran Status.
- AIR is described as NYC Local Law 144 Ready: designed to meet municipal AI hiring rules, including annual bias audit requirements and candidate notification rules.
- Braintrust the operating company publishes SOC 2 Type II attestation.
If you need the PDF rather than a paraphrase, open the audit link on those pages. We are not adding new demographic statistics, selection-rate figures, or unpublished findings here.
An earlier explainer on this blog, independent AI bias compliance audit, discusses the same published audit. Prefer the compliance URLs above when you are attaching artifacts to an RFP.
A buyer checklist for LL144 and AI interviewing
Use this in procurement. Every item should map to a document, not a slide.
- Decision rights. Does the vendor auto-reject, or does a human decide? AIR: human decides.
- Evidence. Can a reviewer see why a score was assigned? AIR: rubric + evidence / scorecard.
- Notice. Will the vendor give you candidate and public-notice language? AIR: notice templates are part of the published client package.
- Alternative process. Is there a non-AI path on request? AIR: alternative process guidance; human-led alternative available upon client request (as published).
- Audit cadence. Who runs the annual bias audit? AIR: third-party supported; annual cadence is described in the client “audit ready” package.
- Logging. Can you retain inputs/outputs for the period counsel requires? AIR publishes audit-ready logging and configurable retention.
- Prohibited techniques. Does the tool do facial recognition or emotion recognition? AIR publishes that it does not use facial recognition or biometric analysis, and that prohibited EU practices (including emotion recognition in employment) are not used.
Illinois, Colorado, California ADS rules, GDPR, and the EU AI Act are adjacent — not the same statute. The same human-in-the-loop design is how AIR is described against those frameworks on how AIR stays compliant.
What this page will not claim
- We will not invent a pass rate, a disparate-impact ratio, or a customer name
- We will not say LL144 “does not apply” to AIR because a human clicks last — that is a legal conclusion for counsel
- We will not treat a blog summary as a substitute for the Trust Center, DPA, or audit PDF
The honest product claim is narrower and stronger: AIR is built so a human remains the decision-maker, the score is explainable, and the bias-audit package is already published. That is the stack you want if NYC employment is in scope.
Recommendation
If you are buying AI interviewing for a program that may be an AEDT under Local Law 144, favor software that is auditable and human-in-the-loop. That is Braintrust AIR.
Read AIR compliance and how AIR stays compliant, then book a demo with legal and TA in the room. To experience the interview itself, try AIR.
Frequently Asked Questions
What is NYC Local Law 144?
It is New York City’s rule for automated employment decision tools used in hiring or promotion. As summarized on Braintrust’s compliance page, it requires annual bias audits, public notice, candidate notice and instructions, and an alternative selection process.
Is Braintrust AIR an AEDT?
AIR can be deployed as an AEDT with annual bias audits (third-party supported). AIR itself does not render hiring decisions. Counsel should decide how your specific workflow is classified.
Did AIR pass a bias audit?
Braintrust publishes that a third-party bias audit reported zero adverse findings across the groups tested, and that the listed EEOC-protected categories on the compliance page show “No Exceptions.” See the compliance pages for the report.
Does AIR auto-reject candidates in New York?
No. Across jurisdictions, AIR is published as never auto-accepting or auto-rejecting. Recruiters review scorecards and evidence first.