NYC Local Law 144 (Automated Employment Decision Tools)

Local Law 144 of 2021 is a New York City ordinance regulating employer use of automated tools in hiring and promotion. It requires an annual independent bias audit of any automated employment decision tool (AEDT), public posting of a summary of the audit results, and advance notice to affected candidates. Enforcement by the Department of Consumer and Worker Protection (DCWP) commenced on July 5, 2023, making it the first US algorithmic-discrimination audit law to take effect.

The ordinance amends the administrative code of the City of New York and is codified at NYC Admin. Code §§ 20-870 to 20-874. It is commonly referred to as "NYC AEDT" or "Local Law 144."

Status and timeline

The New York City Council passed the bill on November 10, 2021, and Mayor Bill de Blasio signed it on December 11, 2021. The original effective date was January 1, 2023, but that date was delayed. The DCWP adopted final implementing rules on April 6, 2023, and enforcement commenced on July 5, 2023. The law was enacted as Local Law 144 of 2021.

Scope and definitions

The law applies to employers and employment agencies using an AEDT for employment decisions concerning a candidate or employee residing in New York City, or for a position located in NYC.

The statute defines an automated employment decision tool as "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."

The DCWP rules interpret "substantially assist or replace" narrowly, covering tools that are (i) the sole criterion for a decision, (ii) weighted more than any other criterion, or (iii) used to override a conclusion derived from other factors.

Key provisions

Bias audit

An independent auditor, defined as someone who is not the developer or deployer of the tool, must conduct the audit. The audit must occur no more than 1 year before use of the AEDT and be updated annually thereafter. It must calculate selection rates and impact ratios for three sets of categories: sex categories, race/ethnicity categories, and intersectional categories (sex × race/ethnicity), the last of which was required under the 2023 final rules. If the employer lacks sufficient historical data, the auditor may use test data, with disclosure.

Public disclosure

The employer must post a summary of the most recent bias audit on the employment section of its website, including the date of the most recent audit, the selection rates and impact ratios by category, and the source and explanation of the data used. The summary must remain posted for at least 6 months after the AEDT ceases to be used.

Candidate notification

The employer must give notice to candidates residing in NYC at least 10 business days before AEDT use. The notice must state three things: that an AEDT will be used, the job qualifications and characteristics the AEDT assesses, and the source and retention of the data collected. A candidate may request an alternative selection process or reasonable accommodation.

Enforcement and penalties

DCWP enforces the law. Civil penalties are $500 for a first violation and $500 to $1,500 per subsequent violation; each day of non-compliance and each affected candidate counts as a separate violation. The law provides no private right of action.

Comparison with other AI legislation

LL 144 is restricted to employment and is the first US algorithmic-discrimination audit law. It predates the Colorado AI Act by roughly two years and served as a practical template for auditor-market formation and impact-ratio methodology, narrower in scope than the broader Colorado deployer-obligation model.

A closer comparison with the Colorado AI Act (SB 24-205) and SB 25B-004 (Date Amendment) highlights the differences:

Reactions and debates

Commentators have identified several contested aspects of the law's design and operation.

The narrow interpretation of "substantially assist or replace" in the final DCWP rules has drawn criticism. Analysts including Deloitte and Brookings argue that the definition lets most modern resume-screening tools escape coverage if the employer claims the tool is merely "one factor" in a decision.

The audit requirement has prompted the creation of a third-party AEDT-audit industry, raising questions of audit quality, conflicts of interest, and methodological standards; the law is silent on auditor qualifications. Dozens of AEDT auditors, including FairNow, Holistic AI, BABL, and Credo AI, have emerged in response.

Critics have characterized enforcement as symbolic, noting that the $1,500-per-violation penalty ceiling is lower than the penalties available under the Colorado AI Act, TRAIGA, or the EU AI Act. The intersectional-reporting requirement has been described as demanding for small employers, who often lack sufficient per-cell population for statistically meaningful intersectional impact ratios, which some argue creates pressure toward discontinuing tool use rather than achieving compliance.

Commentators have also noted that LL 144 regulates disclosure of disparate impact rather than causation: an employer can lawfully use a tool that produces disparate impact provided disclosure is made and no separate challenge follows under other employment-discrimination law, such as Title VII, the New York State Human Rights Law (NYSHRL), or the New York City Human Rights Law (NYCHRL). Because employment discrimination is heavily regulated at the federal level through EEOC guidance, observers have raised implicit-preemption concerns for state and local AEDT laws. Employers also often lack demographic data, particularly on race, for applicants, forcing reliance on test data and statistical proxies.

The law informed the drafting of the Colorado AI Act and has been referenced in EU AI Act implementation guidance on employment systems.

On December 2, 2025, the New York State Comptroller published an audit of the law's implementation that found enforcement gaps: few employers had posted bias audits, and DCWP had issued minimal penalties.

Relationships

• predecessor-to: Colorado AI Act (SB 24-205) and SB 25B-004 (Date Amendment) — direct template for the CO duty-of-care model
• contrasts-with: Texas Responsible AI Governance Act (TRAIGA / HB 149) — Source Summary — intent-based discrimination standard vs. impact-based
• related: EU AI Act (Regulation 2024/1689) — employment AI is Annex III "high-risk" in the EU
• related: US AI Regulatory Approaches Compared — adds municipal-level dimension
• related: Techno-Federalism: How Regulatory Fragmentation Shapes the U.S.-China AI Race — a city-level entrant in the US AI-regulatory stack

Sources

• NYC Admin. Code §§ 20-870–20-874 (enacted LL 144/2021)
• DCWP final rules (Apr 6, 2023): https://rules.cityofnewyork.us/rule/automated-employment-decision-tools-2/
• Secondary: Deloitte (2023); FairNow compliance guide (2026); DCWP AEDT FAQ; NY State Comptroller audit (Dec 2, 2025); Brennan Center / Upturn analysis