New York City's Local Law 144 was the first US law to mandate published third-party bias audits of an AI use case. It governs Automated Employment Decision Tools (AEDTs) used to substantially assist or replace discretionary hiring or promotion decisions for jobs based in NYC.
The law has been operationally in effect since 2023. Enforcement is active. And the template it set, published independent audits, candidate notice, regulator oversight, is being studied and adapted in other jurisdictions. This article walks through what compliance actually looks like.
What counts as an AEDT under the law
An Automated Employment Decision Tool is computational software that substantially assists or replaces discretionary decision-making about hiring or promotion. The key threshold is "substantially assist or replace", tools that merely provide one input among many to a human decision-maker are generally not in scope; tools that drive or strongly influence the decision are.
Tools commonly in scope include, resume screening AI, video interview analysis tools, AI-powered assessment platforms, automated candidate ranking systems, AI-driven sourcing platforms that score and prioritize candidates. Tools not in scope generally include, ATS systems that merely organize applications, scheduling tools, and similar workflow infrastructure that does not assess candidates.
The annual bias audit requirement
Employers using AEDTs must commission an annual independent bias audit. The audit must be performed by a third party that is not the employer, not the AEDT vendor, and not otherwise affiliated. The audit results must be published publicly on the employer's website.
The audit measures impact ratios across protected categories (race, ethnicity, sex) using metrics defined by the law. The specific methodology has been clarified through Department of Consumer and Worker Protection guidance, and qualified audit vendors have established methodologies that produce comparable results.
The published results must include the date of the audit, the AEDT used, the data used in the audit, and the impact ratios calculated. The publication makes the audit findings accessible to candidates, regulators, and the general public, which is itself a deliberate accountability mechanism.
Candidate notice obligations
Employers must notify candidates that an AEDT is being used at least 10 business days before its use. The notice must include, that an AEDT is being used, the job qualifications and characteristics the AEDT assesses, and information about the type and source of data used by the AEDT.
Candidates can request information about the data used and may request an alternative selection process. The alternative request does not have to be granted in every case, but the request mechanism must be available and the response must be documented.
The operational workflow
A compliant Local Law 144 program involves several interlocking workflows:
● AEDT inventory, every AI tool used in hiring or promotion for NYC-based positions, with documentation of what it does and how substantially it influences decisions
● Audit vendor selection and engagement, qualified third-party auditors, with appropriate independence and methodology
● Annual audit execution, providing the auditor with the data they need, supporting their methodology, and receiving the deliverable
● Public posting, audit results published on the employer's website in the format prescribed
● Candidate notice mechanism, notice delivered at least 10 business days before AEDT use, in accessible format
● Alternative process request handling, receiving requests, documenting responses, retaining records
Common implementation pitfalls
● Underscoping the AEDT inventory, only counting tools labeled "AI" rather than any tool that substantially assists or replaces decisions
● Choosing audit vendors based on price rather than methodology rigor, leading to published audits with weak methodology that don't actually demonstrate compliance
● Burying the audit results on a page candidates and regulators can't find
● Notice delivered after AEDT use begins, rather than 10 business days before
● No documented process for handling alternative selection requests, leading to inconsistent responses and potential discrimination claims
● Treating Local Law 144 as a one-time setup rather than an ongoing annual cycle
How Local Law 144 connects to broader hiring AI governance
Local Law 144 is one of the most prescriptive AI hiring laws in the US, but it is not the only applicable regime. Title VII still applies. The EEOC's technical guidance still applies. The ADA still applies. Other jurisdictions (Illinois with AI video interview disclosure, Maryland for facial recognition in hiring, and increasingly other cities and states) have their own hiring AI rules.
A compliant US-wide hiring AI program treats Local Law 144 as the strictest applicable framework and operationalizes against it, published bias audits, transparent notice, alternative process availability, across all hiring AI globally, not just for NYC-based jobs. This is operationally simpler than maintaining city-by-city variants, and it creates a credible compliance story when regulators or candidates raise concerns.
The shift to make
Stop treating NYC Local Law 144 as an NYC-only checklist.
Start treating it as the operational template for the published-audit hiring AI accountability model that is spreading across US jurisdictions. Building genuine compliance, real bias audits with rigorous methodology, accessible notice, working alternative processes, produces a hiring AI program that is defensible everywhere, not just in NYC.
Enterprises that take this seriously gain a quiet competitive advantage in talent attraction, candidates increasingly value AI transparency in hiring, and a structural advantage in regulatory posture as more jurisdictions adopt similar frameworks.
