NYC Local Law 144

NYC Local Law 144 (formally, New York City Local Law 144 of 2021) is a municipal employment law that prohibits employers and employment agencies from using an Automated Employment Decision Tool (AEDT) for hiring or promotion decisions involving New York City candidates or employees unless the tool has been subjected to a bias audit within the preceding 12 months, the audit results have been publicly posted, and the affected candidates or employees have been notified. The New York City Department of Consumer and Worker Protection (NYC DCWP) adopted final implementing rules on April 6, 2023 (codified at Title 6 of the Rules of the City of New York). Enforcement began July 5, 2023.

The law is the first major municipal regulation in the United States specifically targeting AI tools in hiring. Its audit-and-disclose framework — test for disparate impact, post the results publicly, tell candidates you’re using the tool — has influenced legislation in Illinois, California, and federal EEOC guidance, making LL144 compliance a useful template for multi-jurisdiction AI-in-hiring governance.

What is an AEDT under NYC Local Law 144?

The law defines an AEDT as any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues a simplified output — a score, classification, recommendation, or ranking — used to substantially assist or replace discretionary decision-making in:

• Screening candidates for employment, or
• Screening employees for promotion.

A tool meets the “substantially assist or replace” threshold if it acts as the sole basis for a decision, serves as the primary factor among several inputs, or its output has the capability to override conclusions reached by a human. The critical test is not whether the decision is labeled “human” — it is whether the human decision-maker relies substantially on the AI output.

What is NOT an AEDT under the DCWP final rule: tools that merely transcribe interview audio, translate documents, schedule interviews, send automated email sequences, filter spam, perform spreadsheet calculations, or otherwise capture or route data without scoring or recommending. The exemption applies only where no simplified output that bears on the hiring decision is produced.

Consult counsel for classification of specific tools where the vendor’s documentation does not clearly indicate whether the tool produces a simplified output that substantially assists hiring decisions.

Who is covered

The law applies to:

• Employers — any person or entity engaged in business in New York City that uses an AEDT to screen candidates or employees. No minimum employee count.
• Employment agencies — agencies operating in NYC that use AEDTs to screen candidates for placement.
• Candidates and employees — any person who is a resident of New York City, or who is screened for a job located in New York City, or who is considered for promotion in a NYC role.

The law does not require that the employer be physically based in NYC. An employer headquartered elsewhere that uses an AEDT to screen NYC-resident applicants for any role is covered.

The three core requirements

1. Annual bias audit

Before using an AEDT — and then annually — the employer must commission a bias audit from an independent auditor. Independence means:

• No financial relationship with the employer or with the AEDT vendor.
• No prior involvement in the development, use, or distribution of the specific AEDT being audited.

The audit must calculate, for each race/ethnicity category and each sex category (and, where data allows, intersectional combinations):

• The selection rate — the proportion of candidates in that category who received a positive outcome (selected for the next stage, recommended for hire, scored above a threshold).
• The impact ratio — the selection rate for each category divided by the selection rate for the most-selected category. An impact ratio below 0.8 (the EEOC’s “four-fifths rule”) indicates potential adverse impact.

The audit must also report the number of individuals whose race/ethnicity and sex are unknown, and therefore excluded from the calculations.

Vendor-supplied audits: The DCWP rules permit an AEDT vendor to conduct or commission the bias audit on behalf of multiple employers using the same tool. An employer may rely on a vendor-supplied audit — but only if the audit covers the specific version and configuration of the tool as deployed by that employer, and was performed by a genuinely independent auditor. A vendor’s generic compliance document covering the product at a high level does not satisfy the requirement if the employer’s use case differs from the audited configuration.

2. Public posting of audit results

Before using the AEDT for NYC hiring decisions, and then within 30 days of each new annual audit, the employer must post on its public-facing website:

• A summary of the bias audit results, including the selection rates and impact ratios for each category.
• The date the AEDT was first used by the employer.
• The date of the most recent bias audit.

The posting must remain publicly accessible for at least six months after the employer’s latest use of the AEDT (per the DCWP final rule). Employers without a publicly accessible website may satisfy this requirement by making the information available to applicants and employees upon request. Note this is distinct from audit-record retention: the underlying audit documentation should be kept for three years (see the compliance workflow below).

3. Candidate and employee notification

Employers must notify candidates at least 10 business days before the AEDT is used to evaluate them. The notification must include:

• A statement that an AEDT will be used and will substantially assist or replace discretionary decision-making in the hiring process.
• The specific job qualifications and characteristics the tool will assess.
• The types of data collected by the tool, the sources of that data, and how long the data will be retained.

Additionally, candidates must be informed that they may request an alternative selection process or reasonable accommodation in place of the AEDT evaluation (where an alternative exists). The employer is not required to provide an alternative — but must inform candidates of the possibility.

How to deliver the notice: The notice may be provided in the job posting, on the employer’s careers website in a clear and conspicuous location, or via direct written communication (email or postal mail) to each affected candidate. A footnote in general terms of service or a buried privacy policy disclosure does not satisfy the requirement.

Timing implication: If an AEDT is used to screen applications after they are submitted, the 10-business-day notice period means the notice must appear in the original job posting or be sent before candidates apply — not after.

Enforcement and penalties

DCWP is the enforcement agency. Enforcement began July 5, 2023 (delayed from the law’s original effective date due to rulemaking). Penalties are:

• First violation: up to $500.
• Each subsequent violation: $500–$1,500 per violation.

Each day of non-compliance is a separate violation. An employer using an AEDT for 30 days without the required audit results posted could face up to $45,000 in penalties for that period.

The New York State Comptroller’s Office published a December 2025 audit finding that DCWP enforcement had been limited, with few formal enforcement actions in the first two years of the law. This does not reduce the legal obligation — it reflects enforcement capacity, not legal exposure.

What LL144 does NOT require

Common misconceptions:

• It does not ban AI in hiring. It requires audits and disclosure, not discontinuation.
• It does not require the AEDT to produce no disparate impact. It requires the audit to be conducted, the results posted, and candidates notified. An audit showing adverse impact below the four-fifths threshold is a legal exposure requiring a business justification, not automatic non-compliance.
• It does not specify a disparate impact threshold. The four-fifths (80%) ratio is the EEOC’s long-standing analytical benchmark; LL144 adopts the same methodology but does not set an explicit pass/fail threshold. Impact ratios that fall below 0.8 warrant scrutiny and documented business justification.
• It does not apply to all automated tools. Scheduling software, ATS workflow routing, and non-scoring automation are outside the law’s scope.

Compliance workflow for recruiting teams

1. Inventory tools. List every AI tool used in the sourcing-to-hire pipeline and classify each as AEDT or non-AEDT based on whether it produces a simplified output that substantially assists hiring decisions.
2. Obtain or commission bias audits. For each AEDT: obtain the vendor’s audit (verify independence and configuration match) or commission an independent audit. Confirm the audit covers the tool as you use it, not a generic product version.
3. Post results publicly. Before using any AEDT for NYC candidates, publish the audit summary on your careers page or public website. Note the first-use date and audit date.
4. Add notification to job postings. For every role where an AEDT will be used, include the required notification in the job posting. Document the notification language as a standard template.
5. Set annual audit calendar. Audits must be repeated annually. TA Ops owns the renewal calendar; it should be tied to tool contract renewal dates.
6. Maintain records. Audit documentation must be retained for three years. Candidate notifications should be retained as evidence of compliance.

The broader regulatory landscape

LL144 is one node in a growing regulatory framework:

• Illinois AI Video Interview Act (AIVIA): requires disclosure and consent before AI-analyzed video interviews for Illinois candidates; requires employers to report demographic data on AI-screened candidates annually.
• California Assembly Bill 2930 (pending, 2026): would impose bias-audit and impact-assessment requirements on automated decision systems, including hiring tools.
• Federal EEOC guidance (2023): the EEOC’s technical assistance on AI and employment discrimination clarifies that Title VII applies to AI tools that produce disparate impact and that employers remain liable for discriminatory outcomes regardless of whether a vendor operated the tool.
• EU AI Act: classifies AI tools used in employment decisions as high-risk, requiring conformity assessment, registration, ongoing monitoring, and transparency obligations. The high-risk employment obligations were originally set to apply from August 2026, but the EU “Digital Omnibus” political agreement reached on May 7, 2026 provisionally deferred them to December 2, 2027 (pending formal adoption by the Council and Parliament).

Recruiting teams operating across jurisdictions should treat LL144 compliance as the floor, not the ceiling, of their AI governance posture. Consult counsel for jurisdiction-specific analysis.

Common pitfalls

Using a vendor audit that covers the wrong configuration. A vendor that audits its standard product using default settings may not have audited the specific model fine-tuned on your organization’s historical hiring data, or the version with screening thresholds set at the level your team uses. The audit must reflect the actual tool in actual deployment.

Notification in the wrong place. Posting a notice on a general “how we use data” page does not constitute the required AEDT-specific, advance-notice notification. The notice must be in the job posting or sent directly to each candidate before the tool is run.

Treating the first-year audit as permanent. The law requires annual re-audit. A 2023 audit is not valid for 2025 use.

Assuming no impact means no exposure. An audit showing impact ratios above 0.8 for all groups does not mean the tool is free of legal exposure under Title VII or analogous state law. LL144’s audit requirement and federal disparate-impact analysis are separate legal frameworks.

Related

• AI policy for recruiting teams — the internal governance framework that operationalizes LL144 compliance
• AI screening bias — how bias enters AI hiring tools and how to audit for it
• AI resume screening — the specific tool category most commonly triggering AEDT obligations
• Eightfold — talent intelligence platform with built-in bias-audit documentation
• HireVue — video interview platform that has published LL144 bias audit results