NYC Local Law 144: Bias Audits for Automated Employment Decision Tools

A NYC Local Law 144 bias audit is a yearly, independent evaluation of any automated tool used to screen or rank candidates for a job in New York City, and it must be finished before the tool touches an applicant. The New York City Department of Consumer and Worker Protection (DCWP) began enforcing Local Law 144 of 2021 on July 5, 2023. Three duties sit on the employer: get the audit done, publish a summary of the results, and notify candidates at least 10 business days before use.

Key Takeaways

• A tool is covered when it uses machine learning, statistical modeling, data analytics or artificial intelligence to help make an employment decision and substantially assists or replaces discretionary decision making.
• An independent auditor must complete the audit within one year before use, calculating selection or scoring rates and impact ratios across sex, race and ethnicity, and intersectional categories.
• A summary of the most recent results and the tool's distribution date must be published publicly.
• NYC residents applying or up for promotion must be told the tool is in use and what it assesses, at least 10 business days beforehand.
• The employer, not the vendor, is responsible for making sure the audit was done.

What Makes a Tool an Automated Employment Decision Tool

Per DCWP guidance, a covered tool is computer-based, uses machine learning, statistical modeling, data analytics or artificial intelligence, helps make an employment decision, and substantially assists or replaces discretionary decision making. All three conditions must hold. These techniques generate a prediction, such as likelihood of success, or a classification by skill set or aptitude.

Employment decision is broader than the final hire, because screening counts at any stage. DCWP draws a boundary at the front of the funnel: scanning a resume bank, conducting outreach, or inviting applications is not covered, since the law reaches the assessment of people who have applied for a specific position or employees up for promotion.

A scored, model-driven ranking sits inside the definition. A structured instrument whose output a human reads and judges may not, and nor does workflow software that only collects and routes responses, as automated reference checks typically do.

When a Job Counts as Being "In the City"

The law reaches covered tools used for a position in New York City: the job location is an NYC office at least part of the time, the job is fully remote but tied to an NYC office, or the employment agency itself is in NYC. A remote role attached to a Manhattan office is in scope even if the candidate never enters the state.

What the Bias Audit Has to Measure

The audit's minimum content is specified: selection or scoring rates and the impact ratio across sex categories, race and ethnicity categories, and intersectional categories. Intersectional analysis is most often skipped, and a tool can look even on sex and on race while failing badly for one combination.

Impact ratio arithmetic will be familiar to anyone who has run a federal adverse impact check under the Uniform Guidelines on Employee Selection Procedures (29 CFR Part 1607). Local Law 144 sits alongside that federal standard, which remains codified though federal enforcement posture shifted in 2026. Categories under 2% of the audit data may be excluded; every other category must be included.

Historical data from actual use of the tool is the default input. Where historical data is too thin for a statistically significant audit, test data may be used, with the summary explaining how it was sourced. Imputed or inferred demographic data cannot be used at all. DCWP sets no numerical significance threshold, leaving that to the auditor. Our walkthrough of How to Audit Recruitment AI Algorithms for Bias covers assembling that data first.

Who Can Serve as an Independent Auditor

An independent auditor exercises objective and impartial judgment, and DCWP keeps no approved list, so judging independence falls to you. Someone is not independent if they work for the employer or agency using the tool or for the vendor that developed or distributes it, if they were ever involved in using, developing or distributing the tool whoever employs them now, or if they hold a direct or material indirect financial interest in either party.

Publishing the Summary and Giving Notice

The published summary must state the date of the most recent audit, the source and explanation of the data used, the number of individuals assessed falling into an unknown category, and the number of applicants or candidates with selection or scoring rates and impact ratios for all categories. You also publish the distribution date, meaning when you began using the tool, on the employment section of your website or behind an active hyperlink.

The notice must reach employees and job candidates who are New York City residents, say a covered tool is in use, identify the job qualifications or characteristics it assesses, and include instructions for requesting a reasonable accommodation under other laws. It goes out at least 10 business days before use, through the job posting, mail or email. For applicants, a website notice is an accepted alternative, need not be position-specific, and starts the clock from its posting date. For promotion candidates it can sit in a written policy.

• Classify each hiring tool as covered or not covered - Timing: Before any NYC role goes live; Practical note: Record the reasoning for every tool, including exclusions
• Independent bias audit completed - Timing: Within the one year before use, then annually; Practical note: An audit older than one year cannot be relied on
• Publish summary of results and distribution date - Timing: Before first use, refreshed with each audit; Practical note: Must include rates and impact ratios for all categories
• Notice to NYC candidates and employees - Timing: At least 10 business days before use; Practical note: Name the qualifications assessed; include accommodation instructions
• Review results against anti-discrimination duties - Timing: On receipt of each audit; Practical note: The audit itself compels no action; federal and state law still do

Why a Vendor's Bias Badge Is Not Your Compliance File

The employer, not the vendor, must make sure a bias audit was done before a covered tool is used. Many vendors commission an audit of their own product. That is a useful purchasing signal, but it is not your compliance file.

A pooled vendor audit supports your use only if you contributed historical data or this is your first use. And a result averaged across many employers can look clean while your own applicant flow and cut score produce a different picture. Keep your own records: the audit you rely on, the published summary as it appears live, the notice text and send date, and your own impact numbers.

Local Law 144 requires the audit but dictates no action based on the result, and a disparity does not excuse continued use. Title VII duties enforced by the US Equal Employment Opportunity Commission still apply, as do New York State and city anti-discrimination law. DCWP enforces the audit, publication and notice duties and refers discrimination claims to the NYC Commission on Human Rights. Penalties are set by the Administrative Code and assessed per violation, so confirm current amounts with counsel rather than trusting trade coverage.

Using the Local Law 144 Standard Outside New York City

Many employers apply this standard everywhere they hire, because the elements match what any disparate impact defense needs: current independent measurement, documented data sources, candidate transparency, and a named owner. Federal posture shifted in June 2026, when a DOJ Office of Legal Counsel opinion concluded the EEOC's disparate impact guidelines are unconstitutional and the EEOC aligned its enforcement. Exposure relocated rather than disappeared: the liability is statutory at 42 U.S.C. 2000e-2(k), private plaintiffs keep an independent right to sue, and fourteen state attorneys general have affirmed their own enforcement. City and state law is now the sharper exposure, so applying this standard everywhere is a stronger call than before. Illinois and Colorado have moved in a similar direction, and building to the stricter rule once beats rebuilding jurisdiction by jurisdiction.

The cheaper move is often to lean less on automated scoring. Structured, job-relevant measurement that a human reviews keeps you further from the covered-tool definition and improves your validation position. Righteo supports that with role-relevant AI skill assessments and a ready-made skill tests library, and our HR and pre-employment resources cover the surrounding process. Righteo does not perform bias audits or certify legal compliance.

This article is general information for US employers and is not legal advice. Confirm your obligations with employment counsel in the states where you hire.

Related Resources

Automated Employment Decision Tools (AEDT)

Automated Employment Decision Tools: Frequently Asked Questions

Automated Employment Decision Tools: Final Rule

Uniform Guidelines on Employee Selection Procedures (29 CFR Part 1607)

US Equal Employment Opportunity Commission

How to Audit Recruitment AI Algorithms for Bias

HR and Pre-Employment Resources and Articles

Frequently Asked Questions

Does Local Law 144 apply to remote jobs?

Yes, when the fully remote role is associated with an office in New York City. The candidate's physical presence is not the test.

How often does a bias audit have to be repeated?

Every year. An audit can be relied on for one year from the date it was conducted, so continued use needs one completed within the prior year.

Can the vendor's own auditor do our bias audit?

A vendor may commission an audit from an independent auditor, and you may rely on it if you contributed historical data or it is your first use. The auditor still cannot work for the vendor or employer, cannot have helped build the tool, and cannot hold a financial interest in either.

What happens if the audit shows a disparity?

Local Law 144 prescribes no specific response. Federal, state and city anti-discrimination law govern what you do next, which usually means justifying the tool's job relatedness or changing how it is used.

Do we have to notify candidates who do not live in New York City?

The notice duty runs to New York City residents. Many employers notify every applicant anyway, because sorting candidates by residence before assessment is harder than issuing one notice.

Does the 2026 federal change make Local Law 144 compliance optional?

No. Local Law 144 is city law, and a shift in federal enforcement posture does not touch it. The audit, publication and notice duties are unchanged, and DCWP still enforces them.

Compliance Comes Down to a Dull File

Local Law 144 compliance comes down to a dull file: a tool inventory with reasoning, a current independent audit, a published summary with real numbers, and dated notices sent 10 business days ahead of use. Start with the inventory, because nothing else can be assessed until you know what is in scope. Then ask every vendor on it whether they will support an audit on your own applicant flow data.