NYC Local Law 144: AI Hiring Bias Audit Law (2026)

Any employer using an automated tool to help screen, rank, or select candidates for a job tied to a New York City office, even a fully remote one, has to follow NYC Local Law 144: get an independent bias audit completed within the prior year, post the results publicly, and give candidates at least 10 business days' notice before using the tool.

Information last verified on 2026-08-14. This article has not yet been reviewed by a licensed lawyer.

Jurisdiction scope: This article addresses New York City's Local Law 144 (the Administrative Code's automated employment decision tool provisions and DCWP's implementing rule), which applies wherever a covered job is tied to an NYC office regardless of where the employer is headquartered. It also compares Local Law 144 to Illinois, Maryland, Texas, and Colorado's AI-hiring laws and to the current federal EEOC guidance landscape. It does not cover New York State's other AI laws, such as the RAISE Act, the deepfake laws, or the AI companion law; see New York AI Laws for those. It also does not re-cover New York's general employee-monitoring notice rules; see New York Employee Monitoring Laws.

This page is written for any employer, anywhere, that uses software to help screen, rank, or select candidates or employees for a role connected to a New York City office, and for any candidate who wants to know what an employer using that kind of tool owes them. It covers what counts as a covered tool, the audit and notice obligations, what happens if an employer does not comply, and how NYC's approach compares to the small group of other states that currently regulate AI in hiring. For the fuller picture of New York's other AI statutes, see New York AI Laws; for the site's overview of AI regulation nationally, see AI Laws and Regulation in the United States.

What Is an Automated Employment Decision Tool?

Local Law 144 defines an automated employment decision tool (AEDT) as any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues a simplified output, such as a score, classification, or recommendation, used to substantially assist or replace discretionary decision-making for hiring or promotion. The New York City Council's enacted bill text puts it this way:

"The term 'automated employment decision tool' means any computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence, that issues simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary decision making for making employment decisions that impact natural persons." (Int 1894-2020, Local Law 144 of 2021, NYC Council.)

DCWP's own guidance breaks the definition into a three-part test: the tool has to (1) use machine learning, statistical modeling, data analytics, or AI, (2) help an employer or employment agency make an employment decision, and (3) substantially assist or replace discretionary human decision-making. DCWP describes covered outputs as "predictions," assessments of a candidate's likely fit or success, and "classifications," groupings by skill set or aptitude. A tool that only automates a purely mechanical step, without scoring, ranking, or otherwise substituting for a human's judgment call, is less likely to meet the third prong, though DCWP has not published a bright-line list of exempt tools.

Does Local Law 144 Apply to You?

The detail most likely to trip up an employer is that Local Law 144 is triggered by where the job is, not where the candidate lives. DCWP's own FAQ guidance states that the law applies when "the job location is an office in NYC, at least part time," or when "the job is fully remote but the location associated with it is an office in NYC." That means an out-of-state or fully remote-first employer with a nominal NYC office tied to a given role is in scope even if no candidate for that role lives in the five boroughs, and even if the employer has no other NYC presence at all.

This job-location framing is DCWP's own operational gloss on the underlying Administrative Code text, which separately describes the candidate-notice duty as running to "candidates and employees residing in New York City." In practice, DCWP's job-location test is the version employers need to plan compliance around, since it is what the regulator itself applies. An employer unsure whether a specific role counts should look at whether the job posting or job description associates the role with an NYC office, not at where any individual applicant happens to live.

The Bias Audit Requirement

An employer or employment agency may not use an AEDT unless it has been the subject of a bias audit conducted within the one year before the tool is used. The audit has to be performed by an independent auditor, defined by DCWP's rule as someone capable of exercising objective and impartial judgment on all issues within the scope of the audit. An auditor is disqualified if it is or was involved in using, developing, or distributing the specific tool, if it has an employment relationship at any point during the audit with the business using the tool or with its AEDT vendor, or if it has a direct or material indirect financial interest in either.

The audit has to calculate selection rates (or, for a scoring tool, scoring rates) and impact ratios separately across sex categories, race/ethnicity categories (drawn from the EEOC's EEO-1 reporting categories), and intersectional categories that combine sex with race and ethnicity, for example comparing the selection rate for Hispanic or Latino male candidates against Black or African American female candidates. That intersectional layer is not optional: DCWP's rule requires it in addition to, not instead of, the separate sex and race/ethnicity comparisons, because a tool can clear both individual comparisons while still showing a disparate impact once the categories are combined. Because the audit has to be redone at least annually to remain current, an employer that keeps using the same AEDT past the one-year mark without a fresh audit falls out of compliance even if nothing else about its use of the tool has changed.

Public Posting and Candidate Notice

Before using an AEDT, an employer has to publicly post, in a clear and conspicuous place on the employment section of its website, the date of the tool's most recent bias audit, a summary of the audit's results, and the AEDT's distribution date. This posting duty exists independently of the notice owed to any individual candidate.

Separately, an employer has to notify candidates that an AEDT will be used at least 10 business days before it is applied, describe how the tool will be used, disclose what job qualifications or characteristics it will assess, and tell candidates they may request an alternative selection process or a reasonable accommodation. On written request, an employer also has to disclose the type of data collected for the AEDT and its source, and how the data was used, within 30 days of the request.

Penalties for Noncompliance

Local Law 144 sets a civil penalty of up to $500 for a first violation, including any additional violation occurring on the same day as the first, and $500 to $1,500 for each violation after that. Each day an employer uses a noncompliant AEDT counts as a separate violation, and each failure to provide a required notice is also its own separate violation, so an employer that keeps using an unaudited tool without fixing its notice practices can accumulate multiple penalties at once. Notably, DCWP's own public-facing guidance materials do not restate these dollar figures; the enacted Administrative Code penalty provision, reflected in the City Council's bill text, is the source for the amounts above.

Enforcement in Practice: What New York's Comptroller Found

The law on the books and the law as enforced have not matched up well so far. A December 2, 2025 audit by the New York State Comptroller, covering DCWP's enforcement from July 2023 through June 2025, the law's first two years, found DCWP's enforcement of Local Law 144 "ineffective."

Over that two-year window, DCWP received only 2 AEDT-related complaints. Of 32 companies the Comptroller's office reviewed, DCWP itself identified just 1 instance of noncompliance, while the Comptroller's auditors identified at least 17 instances of potential noncompliance among that same group of 32 companies. The audit does not report any penalties as having been issued during the review period.

The Comptroller's criticisms centered on process: DCWP relied on what the audit called an ineffective complaint process, did not use enforcement procedures it had jointly developed with the city's Office of Technology and Innovation (OTI), did not draw on OTI's technical expertise when evaluating AEDTs, and did not address every Local Law 144 requirement in the reviews it did conduct. The Comptroller recommended DCWP fix its complaint routing, investigate why so few complaints were filed, shift toward proactive rather than purely complaint-driven enforcement, use OTI's enforcement workbook, and cover the law's full set of requirements in future reviews.

The practical takeaway for an employer is not that Local Law 144 can safely be ignored. The audit and notice obligations remain legally binding regardless of how often DCWP has enforced them so far, and the Comptroller's findings are themselves a call for DCWP to enforce more aggressively going forward, not evidence that it will keep enforcing lightly.

How NYC Compares to Other States' AI-Hiring Laws

NYC's audit-and-disclosure model is one of the more demanding approaches on the books, but it is not the only design other jurisdictions have chosen.

Illinois's HB 3773 is a useful case study in how unsettled this area still is. The Illinois Department of Human Rights published proposed implementing rules on May 15, 2026, then withdrew them on June 2, 2026, citing a need for continued coordination with other state agencies, and postponed a June 10, 2026 public hearing without setting a new date. The underlying January 1, 2026 statutory effective date was not delayed, only the implementing rules were, so Illinois employers already have a live discriminatory-effect prohibition and notice duty in force even though the exact form of the required notice remains unsettled. Maryland's HB 1202 is worth noting mainly for how much narrower it is than the others: it restricts creating a facial recognition template during an interview unless the applicant signs a written waiver, and does not reach general automated hiring-decision tools or require any audit. Texas's TRAIGA takes close to the opposite design choice from NYC and Illinois, requiring intentional discrimination and excluding disparate-impact claims, with no mandated bias audit or candidate-disclosure duty and enforcement limited to the state attorney general; consult the statute directly, or the site's Texas AI Laws page, before relying on the specifics of a private right of action. Colorado's original 2024 AI Act, SB 24-205, would have imposed risk-management programs, impact assessments, and a general duty of reasonable care to prevent algorithmic discrimination, but the legislature repealed and replaced it with SB 26-189 in May 2026 before any of that ever took effect. The law actually scheduled to take effect January 1, 2027 drops those obligations for a narrower automated-decision-making-technology (ADMT) regime: notice before use, a plain-language explanation within 30 days of an adverse decision, a right to correct inaccurate data, and a right to request human review. Like NYC and Illinois, Colorado's incoming law is a disclosure-and-rights framework rather than a pure intent-based standard; unlike them, it still does not require an independent bias audit. See Colorado AI Laws for the current requirements ahead of the 2027 effective date.

No comprehensive federal AI-hiring statute exists. The EEOC's 2022 guidance, "The Americans with Disabilities Act and the Use of Software, Algorithms, and Artificial Intelligence to Assess Job Applicants and Employees," remains posted on eeoc.gov with no withdrawal notice. By contrast, the EEOC's May 2023 Title VII technical assistance document, "Select Issues: Assessing Adverse Impact in Software, Algorithms, and Artificial Intelligence Used in Employment Selection Procedures," has been taken down; its URL returns a 404, even though the same title still appears, still linking to that dead URL, on the EEOC's own publications index. That gap between an active ADA-focused document and a missing Title VII-focused one is a fair illustration of how unsettled federal guidance on AI in hiring currently is; employers should not assume either document reflects the EEOC's complete current position.

California's privacy regulator, the California Privacy Protection Agency (CPPA), has also moved in this space: its automated decision-making technology (ADMT) regulations were approved by the Office of Administrative Law on September 22, 2025 and took effect January 1, 2026, and cover "significant decisions," a category legal commentary describes as including employment and independent-contracting opportunities and compensation. Some law-firm summaries describe a phased compliance deadline of January 1, 2027 for systems already in use; that later date was not independently confirmed against the regulation text itself for this article, so an employer relying on it for compliance planning should confirm the current deadline directly with the CPPA or its own counsel rather than relying on this page alone.

One more federal development cuts the other way. A December 11, 2025 executive order, Ensuring a National Policy Framework for Artificial Intelligence (EO 14365), directs a Department of Justice AI Litigation Task Force to challenge state and local AI laws in court on preemption and constitutional grounds, and DOJ had already intervened in litigation over Colorado's AI law by mid-2026. No court has invalidated Local Law 144 or any comparable AI-hiring law as of this writing, and the executive order itself does not suspend any of them. But the litigation is ongoing, so employers should treat the current state and local compliance landscape as live law that is under active federal challenge rather than a settled long-term framework.

Related Resources

• AI Laws and Regulation in the United States covers the nationwide landscape of AI regulation beyond hiring, including the states and federal proposals not detailed here.
• New York AI Laws covers New York State's other AI statutes, including the RAISE Act, the AI companion law, and the state's deepfake laws.
• New York Employee Monitoring Laws covers New York's general workplace monitoring notice rules, a separate topic from AI-driven hiring decisions.
• Illinois AI Laws covers the Artificial Intelligence Video Interview Act and Public Act 103-0804 in more depth.
• Maryland AI Laws covers HB 1202's facial-recognition-in-interviews restriction.
• Texas AI Laws covers TRAIGA's intent-based, audit-free approach.
• Colorado AI Laws covers the Colorado AI Act's automated decision-making technology (ADMT) framework ahead of its January 1, 2027 effective date.

---

Last updated: 2026-08-14.