NYC Local Law 144: What It Is, Who It Applies To, and What Employers Must Do

NYC Local Law 144 is a New York City law regulating the use of automated employment decision tools by employers and employment agencies operating in New York City. It came into effect on 5 July 2023 following a one-year enforcement delay and requires employers using AI or algorithmic tools to make or assist employment decisions to conduct independent bias audits, publish audit results, and notify candidates and employees that such tools are being used.

Local Law 144 is the first binding municipal AI employment regulation in the United States and has served as a model for subsequent state-level AI employment legislation including the Colorado AI Act and proposed federal frameworks. For non-US organisations, it represents the leading example of jurisdiction-specific AI employment regulation operating alongside broader frameworks such as the EU AI Act.

This guide explains what Local Law 144 requires, who it applies to, what an automated employment decision tool is under the law, what the bias audit must cover, what notice must be given, what the penalties are, and how Local Law 144 interacts with other AI regulation including the EU AI Act.

Key Definitions

Term	Definition
Automated employment decision tool (AEDT)	Any computational process derived from machine learning, statistical modelling, 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 in employment decisions
Bias audit	An impartial evaluation of an AEDT by an independent auditor to assess the tool’s disparate impact on persons based on sex, race, and ethnicity
Employment decision	Decisions about hiring or promotion of candidates and employees
Independent auditor	A person or organisation that is not employed by, and has no financial interest in, the employer, employment agency, or AEDT vendor being audited
Disparate impact	A statistically significant difference in selection rates between groups defined by sex, race, or ethnicity
Selection rate	The rate at which candidates or employees are selected for advancement to the next stage of the hiring or promotion process, or for hire or promotion
Employer	Any person or entity employing four or more employees, where those employees work in New York City
Employment agency	Any person or entity undertaking to procure employees or opportunities for employment

What Is NYC Local Law 144

Local Law 144 of 2021, signed into law by the Mayor of New York City on 11 November 2021, amended the New York City Administrative Code to regulate the use of automated employment decision tools. After a one-year delay requested by the employment technology industry, enforcement began on 5 July 2023.

The law was enacted in response to documented evidence that AI and algorithmic hiring tools, including CV screening software, video interview analysis tools, and candidate scoring platforms, can produce discriminatory outcomes by encoding historical hiring patterns that disadvantage women, people of colour, and other protected groups. The New York City Council concluded that existing anti-discrimination law under the New York City Human Rights Law was insufficient on its own to address algorithmic discrimination, because traditional discrimination law requires proving discriminatory intent or identifying a specific discriminatory practice, whereas algorithmic bias can arise from facially neutral processes that produce disparate outcomes.

Local Law 144 addresses this through a proactive bias audit requirement: employers must assess whether their AEDTs produce disparate impacts before harm occurs, rather than waiting for individual complainants to bring discrimination claims after the fact.

Who Local Law 144 Applies To

Local Law 144 applies to employers and employment agencies that use an AEDT to make employment decisions concerning positions that will be performed, at least in part, in New York City.

Employers

An employer covered by Local Law 144 is any person or entity employing four or more employees where those employees perform work, at least in part, in New York City. The four-employee threshold tracks the threshold for coverage under the New York City Human Rights Law.

There is no revenue threshold, industry threshold, or establishment threshold. A company headquartered in California using an AI hiring tool for a New York City role is within scope. A company established in London using the same tool for a New York City position is within scope. Territorial scope follows the location of the work, not the location of the employer.

Employment Agencies

Employment agencies that use AEDTs to screen candidates for positions to be performed in New York City are also within scope. This includes staffing firms, executive search firms, and recruitment process outsourcing providers that use algorithmic tools to rank, filter, or recommend candidates.

Who Is Not Covered

Local Law 144 does not apply to employers with fewer than four employees. It does not apply to positions performed entirely outside New York City. It does not apply where the employer uses no AEDT in the employment decision, regardless of how much data processing is involved in the hiring process.

The law applies only to hiring and promotion decisions. It does not cover performance management, termination, compensation, or scheduling decisions, even where those decisions are made using algorithmic tools.

What Is an Automated Employment Decision Tool

The definition of an automated employment decision tool is the most litigated and contested element of Local Law 144. The New York City Department of Consumer and Worker Protection (DCWP), which enforces the law, has issued rules and guidance clarifying the definition, but ambiguity remains in edge cases.

The Statutory Definition

An AEDT is any computational process derived from machine learning, statistical modelling, 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 employment decisions.

Four elements of this definition are operative.

Computational process derived from machine learning, statistical modelling, data analytics, or artificial intelligence. This captures AI-powered hiring tools, algorithmic scoring systems, and statistical models used in employment contexts. It does not capture purely rule-based systems where human beings have explicitly programmed the decision rules without any machine learning or statistical derivation.

Issues simplified output including a score, classification, or recommendation. The tool must produce an output that simplifies the decision for the human decision-maker: a ranking, a pass/fail classification, a recommended shortlist, or a numeric score. A system that merely presents raw data to a human decision-maker without generating a simplified output may not satisfy this element.

Used to substantially assist or replace discretionary decision-making. The tool must be used in the actual employment decision, not merely as background research or for administrative purposes unconnected to the decision. A tool is used to substantially assist if the output is a principle basis for the employment decision. A tool is used to replace discretionary decision-making if the employer relies on the tool’s output without meaningful independent human review.

For employment decisions. The tool must be used in hiring or promotion decisions. Other employment decisions are not within scope.

What Counts as an AEDT in Practice

Tool type	AEDT under Local Law 144
AI CV screening software ranking candidates	Yes
Video interview analysis tools scoring body language or speech patterns	Yes
Candidate scoring platforms generating hire/no-hire recommendations	Yes
Automated skills assessment tools generating ranked scores	Yes
Applicant tracking system (ATS) used only for scheduling and record-keeping	No
Rule-based keyword filter programmed by humans without ML	Depends on DCWP guidance interpretation
LinkedIn recruiter search used by a human to find candidates	No
Reference checking platform generating risk scores	Yes
Salary benchmarking tool not used in hiring decisions	No
Performance management AI not used in promotion decisions	No

The Bias Audit Requirement

What a Bias Audit Must Cover

The bias audit is the centrepiece of Local Law 144. It must be conducted by an independent auditor before the AEDT is used and at least annually thereafter. The audit must assess the tool’s disparate impact on employment candidates or employees based on sex, race, and ethnicity.

The DCWP’s rules specify that the bias audit must calculate the selection rate for each category of sex, race, and ethnicity and compute the impact ratio for each category relative to the most selected category. The impact ratio is the standard disparate impact metric: the selection rate of a given category divided by the selection rate of the category with the highest selection rate.

Calculation	Formula
Selection rate (group X)	Number of persons in group X selected / Total number of persons in group X who applied the tool to
Impact ratio (group X)	Selection rate of group X / Selection rate of highest-selected group
Disparate impact threshold	An impact ratio below 0.8 (the four-fifths rule) indicates significant disparate impact

The four-fifths rule, also known as the 80% rule, is the EEOC’s standard for assessing adverse impact in employment selection. Local Law 144 incorporates this standard by requiring the impact ratio to be calculated and published. It does not set a mandatory minimum impact ratio: an employer is not automatically prohibited from using an AEDT with an impact ratio below 0.8, but they must publish the result and may face heightened scrutiny under the New York City Human Rights Law.

Who Can Conduct the Bias Audit

The auditor must be independent: a person or organisation that is not employed by the employer, the employment agency, or the AEDT vendor, and that has no financial interest in the employer, employment agency, or vendor being audited.

In practice, this means the employer cannot use its internal data science team to conduct the bias audit, and the AEDT vendor cannot audit its own tool. The audit must be conducted by a third-party organisation with no commercial relationship that could compromise independence.

The DCWP’s rules do not require the independent auditor to hold any specific professional certification, but the auditor must have sufficient expertise to conduct the required statistical analysis.

Data Requirements for the Audit

The bias audit must be conducted using historical data from the AEDT’s use by the employer or employment agency, where such data is available. Where the employer has used the AEDT for fewer than six months, the employer may use data from the AEDT vendor’s other clients, provided that data is not specific to those clients’ candidates.

The DCWP’s rules specify minimum data requirements. The audit dataset must include at least one year of data where available. Where intersectional categories are included in the analysis, the auditor must note where categories have insufficient sample sizes.

Data source	When it can be used
Employer’s own historical AEDT data	Preferred. Used when employer has sufficient historical data
Vendor test data from other clients	Used when employer has used AEDT for fewer than six months or lacks sufficient data
Combined employer and vendor data	Permissible where employer data is supplemented by vendor data

Publication Requirements

Local Law 144 requires employers to publish a summary of the bias audit results on their website at least ten business days before using the AEDT for a covered employment decision.

The published summary must include the date of the most recent bias audit, the source and explanation of data used to conduct the audit, the number of individuals the audit data encompassed, the number and percentage of individuals falling within an unknown category, and the impact ratios for all categories of sex, race, and ethnicity assessed.

The published summary must remain on the employer’s website for at least six months after the employer ceases to use the AEDT.

Publication requirement	What must be included
Audit date	Date the most recent bias audit was conducted
Data source	Description of data used and its source
Sample size	Number of individuals in the audit dataset
Unknown category	Number and percentage of individuals in unknown demographic categories
Impact ratios	Impact ratios for all sex, race, and ethnicity categories assessed
Publication timing	At least 10 business days before AEDT is used
Retention period	At least 6 months after AEDT use ceases

What Are the Notice Requirements?

Employers must notify candidates and employees that an AEDT will be used in connection with their assessment before the tool is used. The notice must be provided at least ten business days before the AEDT is applied to the individual.

The notice must state that an AEDT will be used, identify the job qualifications and characteristics the AEDT will use to assess the candidate, and provide information about the types of data collected for the AEDT and how long it will be retained.

Candidates must also be given the opportunity to request an alternative selection process or accommodation. The employer is not required to offer an alternative process, but must inform candidates that they can make the request.

Notice requirement	What must be provided
Timing	At least 10 business days before AEDT is applied
Content: AEDT disclosure	Statement that an AEDT will be used
Content: assessment criteria	Job qualifications and characteristics the AEDT will assess
Content: data collected	Types of data the AEDT collects and retention period
Content: accommodation	Information that candidates may request an alternative process
Delivery method	Direct notice to candidate or employee, or posted notice for NYC-based positions

Where the employer does not have the candidate’s contact information, such as where applications are submitted through a third-party platform, the notice may be posted on the employer’s website or the relevant job posting page.

What are the Penalties and Enforcement?

Local Law 144 is enforced by the New York City Department of Consumer and Worker Protection. The DCWP has authority to investigate complaints, conduct audits of employer compliance, and impose civil penalties.

Violation	Civil penalty
First violation	Up to USD 375 per day the violation continues
Second and subsequent violations	Up to USD 1,500 per day the violation continues
Failure to conduct bias audit	Up to USD 1,500 per day, calculated from the date the AEDT was first used without a compliant audit
Failure to publish audit results	Up to USD 1,500 per day, calculated from the date publication was required
Failure to provide candidate notice	Up to USD 1,500 per day per affected candidate

Each day of non-compliance is a separate violation. An employer that fails to conduct a bias audit for a full year before using an AEDT faces potential penalties accumulating over the entire period of non-compliance. The DCWP has indicated it will consider the nature of the violation, the employer’s size, and whether the employer took good-faith remediation steps in determining penalty amounts.

The DCWP can also refer cases to the New York City Commission on Human Rights where AEDT use appears to violate the New York City Human Rights Law. Human Rights Law violations carry separate and more significant remedies including compensatory damages, civil penalties of up to USD 250,000 for wilful violations, and injunctive relief.

Local Law 144 and the EU AI Act

Local Law 144 and the EU AI Act address overlapping territory from different regulatory starting points. Both regulate AI use in employment decisions. Both impose transparency requirements. Both require some form of impact assessment. The compliance obligations are distinct and both apply to multinational employers with New York City and EU operations.

Feature	NYC Local Law 144	EU AI Act (Annex III, employment)
Scope	Employment decisions for NYC positions	Employment decisions affecting EU residents
Risk classification	All AEDTs covered equally	AI Act classifies employment AI as high-risk
Bias assessment	Mandatory bias audit by independent auditor	Data governance under Article 10 includes bias examination
Transparency	Published audit results	Technical documentation under Annex IV
Individual notice	Required before AEDT applied	Deployer must inform individuals under Article 26(8)
Human oversight	Not explicitly required	Required under Article 14
Fundamental Rights Impact Assessment	Not required	Required for certain deployers under Article 27
Enforcement	DCWP civil penalties	National market surveillance authorities, fines up to €15M or 3% turnover
Effective date	5 July 2023	December 2027 (Annex III deadline)

A multinational employer using an AI hiring tool for positions in both New York City and EU member states must satisfy both frameworks simultaneously. The Local Law 144 bias audit does not substitute for the EU AI Act’s data governance, technical documentation, and FRIA requirements. The EU AI Act’s FRIA does not substitute for the Local Law 144 bias audit. Both must be completed independently and documented separately.

Practical Overlap

Despite being separate requirements, the Local Law 144 bias audit and the EU AI Act’s Article 10 data governance and Article 27 FRIA assessment cover overlapping analytical ground. An employer conducting a bias audit for Local Law 144 compliance will generate data on selection rate disparities by demographic group that is directly relevant to the EU AI Act’s fundamental rights impact analysis. Structuring the bias audit to generate outputs usable in both frameworks reduces duplication of effort.

The AI Act’s transparency requirements under Article 26(8) require deployers to inform affected individuals that they are subject to a high-risk AI system. Local Law 144’s notice requirements require employers to inform candidates that an AEDT will be used. The content requirements differ but both can be addressed through a single candidate-facing disclosure designed to satisfy both standards simultaneously.

Interaction with Other US AI Employment Laws

Local Law 144 operates within a growing landscape of US AI employment regulation. Employers with multi-state operations must track requirements across several jurisdictions.

Jurisdiction	Instrument	Status	Key requirement
New York City	Local Law 144	In force since July 2023	Bias audit, publication, candidate notice
Colorado	Colorado AI Act	In force from February 2026	Algorithmic impact assessment, transparency, consumer rights
Illinois	AI Video Interview Act	In force since January 2020	Notice and consent for AI video analysis
Maryland	MD HB1202	In force since October 2020	Consent for facial recognition in employment
New York State	Various proposed bills	Not yet adopted	Proposed bias audit requirements at state level
Federal (US)	No binding federal AI employment law	Voluntary frameworks only	EEOC guidance on AI and employment discrimination

The EEOC’s technical assistance on AI and the Uniform Guidelines on Employee Selection Procedures apply to employment AI tools under existing federal anti-discrimination law. These do not create standalone audit requirements but establish the legal context within which Local Law 144 penalties and Human Rights Law claims operate.

Your Key Takeaway: Compliance Checklist

Requirement	Timing	Status
Identify all AEDTs used in NYC employment decisions	Before any AEDT use	☐
Engage independent auditor with no financial interest in employer or vendor	Before bias audit	☐
Conduct bias audit covering sex, race, and ethnicity impact ratios	Before AEDT use and annually thereafter	☐
Publish bias audit summary on employer website	At least 10 business days before AEDT use	☐
Ensure published summary includes all required elements	At publication	☐
Retain published summary for 6 months after AEDT use ceases	Ongoing	☐
Provide candidate notice at least 10 business days before AEDT applied	Before each AEDT application	☐
Include all required notice content: AEDT disclosure, assessment criteria, data collected, accommodation right	At notice	☐
Document bias audit process and results	Ongoing	☐
Conduct annual re-audit	Annually	☐
Assess Local Law 144 compliance alongside EU AI Act obligations where applicable	Before EU and NYC positions are filled using AEDT	☐

How does Grecta support Local Law 144 compliance for organisations also subject to the EU AI Act?

Grecta’s AI governance platform operates at the product and system level, mapping AI system obligations across multiple regulatory frameworks simultaneously. For multinational employers using AI hiring tools subject to both Local Law 144 and the EU AI Act, Grecta supports the classification, documentation, and evidence management processes required by both instruments, reducing the duplication of compliance work across jurisdictions.

FAQ