For employers
NYC Local Law 144: when does an AI hiring tool need a bias audit, and what must the employer publish and tell candidates?
Since January 1, 2023, New York City has barred employers and employment agencies from using an automated employment decision tool to screen candidates or employees unless the tool had a bias audit within the past year and a summary of it is public. Candidates must be told at least ten business days in advance. A December 2025 state audit counted two complaints in two years and at least 17 potential violations at 32 companies where the city found one.

What does NYC Local Law 144 require of employers?
A bias audit of the tool by an independent auditor no more than one year before use, a public summary of its results on the employer's website before use, and notices to candidates and employees who reside in the city.
Local Law 144 of 2021 added sections 20-870 to 20-874 to the New York City Administrative Code. Section 20-871(a) makes it unlawful for an employer or an employment agency to use an automated employment decision tool "to screen a candidate or employee for an employment decision unless" two conditions are met.
The first condition is a bias audit "conducted no more than one year prior to the use of such tool." The second is that a summary of the results of the latest audit, with "the distribution date of the tool to which such audit applies," "has been made publicly available on the website of the employer or employment agency prior to the use of such tool."
The law took effect on January 1, 2023. The Department of Consumer and Worker Protection (DCWP) adopted rules in Subchapter T of Chapter 5 of Title 6 of the Rules of the City of New York, sections 5-300 to 5-304, and enforcement began on July 5, 2023.
Which hiring tools count as an automated employment decision tool?
A computational process derived from machine learning, statistical modeling, data analytics or artificial intelligence that issues a simplified output, such as a score or a ranking, used to substantially assist or replace discretionary decisions on hiring or promotion.
Section 20-870 defines an automated employment decision tool as a "computational process, derived from machine learning, statistical modeling, data analytics, or artificial intelligence," that "issues simplified output, including a score, classification, or recommendation," which is "used to substantially assist or replace discretionary decision making for making employment decisions that impact natural persons." An employment decision means "to screen candidates for employment or employees for promotion within the city."
The definition excludes tools that do not automate or replace such decisions, and it names examples: "a junk email filter, firewall, antivirus software, calculator, spreadsheet, database, data set" and other compilations of data.
In practice the definition reaches resume screeners that rank applicants, video interview systems that score answers, and online assessments whose scores decide who moves to the next round. The question for each tool is whether its output substantially assists or replaces the decision that a person would otherwise make.
What is a bias audit, and who may conduct it?
An impartial evaluation by an independent auditor that calculates selection or scoring rates and impact ratios by sex, by race or ethnicity, and for intersectional categories. The auditor may not have used, developed or distributed the tool.
Section 20-870 defines a bias audit as "an impartial evaluation by an independent auditor." Rule 5-300 states that an independent auditor is a person or group "capable of exercising objective and impartial judgment on all issues within the scope of a bias audit of an AEDT," and that an auditor is not independent if it "is or was involved in using, developing, or distributing the AEDT."
Rule 5-301 requires the audit to calculate the selection rate or the scoring rate and the impact ratio for each category, separately for sex categories, race and ethnicity categories, and intersectional categories that combine the two. Its example is the "impact ratio for selection of male candidates vs female candidates." Under rule 5-301(d), the auditor "may exclude a category that represents less than 2% of the data being used for the bias audit."
Rule 5-302 states that the audit "must use historical data of the AEDT." Test data may be used only "If insufficient historical data is available to conduct a statistically significant bias audit."
How is the impact ratio calculated?
The selection rate of a category divided by the selection rate of the category selected at the highest rate, or, for tools that score, the scoring rate of a category divided by that of the highest scoring category.
Rule 5-300 defines the impact ratio in two forms. The first is the selection rate for a category divided by the selection rate of the category selected at the highest rate. The second, for tools that score, is "the scoring rate for a category divided by the scoring rate for the highest scoring category."
The selection rate is the share of a category that moves forward or receives a classification. The scoring rate is "the rate at which individuals in a category receive a score above the sample's median score, where the score has been calculated by an AEDT."
A worked example with invented figures: a screener advances 120 of 400 male applicants and 90 of 400 female applicants. The selection rates are 30% and 22.5%. Men are selected at the higher rate, so the impact ratio for women is 22.5 divided by 30, or 0.75, and the ratio for men is 1.0. The summary reports both numbers.
Does the law ban a tool with a low impact ratio?
No. DCWP's guidance says the law does not require any specific action based on audit results. Federal anti-discrimination law still applies, and the four-fifths rule of the Uniform Guidelines treats a ratio below 0.8 as evidence of adverse impact.
The DCWP frequently asked questions state that "the Law does not require any specific actions based on the results of a bias audit," and that employers "must comply with all relevant Anti-Discrimination laws and rules" when deciding what to do after an audit.
Under 29 CFR 1607.4(D), "A selection rate for any race, sex, or ethnic group which is less than four-fifths (4/5) (or eighty percent) of the rate for the group with the highest rate will generally be regarded by the Federal enforcement agencies as evidence of adverse impact." The impact ratio in a Local Law 144 summary is the same comparison, so the ratio of 0.75 in the example above falls below that line.
The Guidelines add that "Smaller differences in selection rate may nevertheless constitute adverse impact, where they are significant in both statistical and practical terms," and that greater differences may not where they rest on small numbers and are not statistically significant.
What must the published audit summary contain?
The date of the latest audit, the source and explanation of the data, the number of people assessed, those in an unknown category, the selection or scoring rates and impact ratios for each category, and the distribution date.
Rule 5-303 lists the contents of the summary posted, in a clear and conspicuous manner, on the "employment section of their website". It includes the date of the latest bias audit, the source and an explanation of the data used, the selection or scoring rates and impact ratios for all categories, and "the number of individuals the AEDT assessed that fall within an unknown category." The distribution date of the tool is posted as well.
Under rule 5-303(c), the employer keeps the summary and the distribution date posted "for at least 6 months after its latest use of the AEDT for an employment decision." A new audit is required before the one-year period runs out, so a tool used for several years produces a series of summaries.
What must candidates and employees be told, and when?
At least ten business days before use: that a tool will be used, which qualifications and characteristics it assesses, and that an alternative process or an accommodation may be requested. Data details follow a written request within 30 days.
Section 20-871(b) requires notice to each candidate or employee who resides in the city "no less than ten business days before such use." The notice states that an automated employment decision tool will be used, tells the candidate that he or she may "request an alternative selection process or accommodation," and lists the job qualifications and characteristics that the tool will use in the assessment.
Information about the type of data collected for the tool, its source and the employer's data retention policy must be available on the website or provided on written request: "Such information shall be provided within 30 days of the written request."
Rule 5-304 lists the ways to give notice: on the employment section of the employer's website, in the job posting, or by U.S. mail or email. Each method has to give the notice at least ten business days before the tool is used.
What are the penalties, and how is the law enforced?
Up to $500 for a first violation and $500 to $1,500 for each later one; each day of unlawful use counts separately. A 2025 state audit found at least 17 potential violations at 32 companies where the city found one.
Section 20-872 sets a civil penalty of "not more than $500 for a first violation" and "not less than $500 nor more than $1,500 for each subsequent violation." "Each day on which an automated employment decision tool is used in violation of this section" is a separate violation, and a failure to give notice to a candidate is a separate violation as well.
The Office of the New York State Comptroller issued audit report 2024-N-6 on December 2, 2025, covering July 2023 through June 2025. DCWP received two complaints about automated tools in that period. DCWP reviewed the websites and bias audits of 32 companies and found one instance of noncompliance; the auditors reviewed the same companies and found at least 17 instances of potential noncompliance.
The Comptroller recommended that DCWP route complaints received through 311 to the agency, look into why so few complaints arrive, review audits against all requirements of the law and identify potential violations through research and other leads in addition to complaints. The report also noted that DCWP officials said they lacked technical expertise to evaluate tool use and had not consulted the city's Office of Technology and Innovation.
Does a skills test scored by people fall under the law?
The definition turns on a computational process whose simplified output substantially assists or replaces the decision. Where a person reads the answers and decides the score, the employer's counsel checks each step against that definition.
Local Law 144 does not regulate tests as such. It regulates computational processes derived from machine learning, statistical modeling, data analytics or artificial intelligence whose output is used to substantially assist or replace discretionary decisions. A written exam scored by an examiner against a rubric, with the decision made by people, is a different arrangement from a system that ranks applicants on its own.
Federal rules apply to selection procedures whether or not they are automated. The four-fifths rule in 29 CFR 1607.4(D) speaks of the selection rate for any race, sex or ethnic group, so an employer that uses a scored exam keeps the records needed to calculate selection rates by group.
What does the IAE written exam measure, and what does the company see?
The IAE exam is an online written exam of up to 2 hours in the employee's field: 8 tasks scored against the field rubric, one of them about the employee's own project. An examiner approves the result.
Each of the 8 tasks has its own timer: 10 minutes per question and 25 minutes per case, taken in one sitting. Answers are typed in the exam form with pasting disabled, the system records how each answer was typed, and each candidate receives their own figures and circumstances in the cases. One question concerns a project from the candidate's questionnaire: the decisions taken, the figures and the outcome.
Answers are scored against the rubric of the field, and an examiner with no ties to the employee or the company approves the result within 5 business days. Only the IAE Secretariat sees the questionnaire on experience, education and the project. The company sees in its cabinet the status, the total score, the rubric criterion scores for each task and the certificate PDF if the exam is passed; it does not see the questionnaire, the CV or the reason for refusal.
A company registers through the page for employers and receives a link to the company cabinet within one business day, with no calls or meetings. In the pilot, companies that register by December 31, 2026 test up to 5 employees free of charge. The employer remains responsible for how it uses any test in its own selection process, including the notices and audits that city or state law requires.
Sources
- New York City Local Law 144 of 2021 with DCWP rules (Subchapter T, 6 RCNY 5-300 to 5-304) and FAQ, copy published by IMLA
- Rules of the City of New York, Title 6, Chapter 5, Subchapter T, Automated Employment Decision Tools, section 5-300 (American Legal Publishing)
- Office of the New York State Comptroller, Enforcement of Local Law 144, Automated Employment Decision Tools, report 2024-N-6, December 2, 2025
- 29 CFR 1607.4, Information on impact (LII, Cornell Law School)
Read also
All articles
Does an employer have to pay for required training, certification exams and study time?
Under the Fair Labor Standards Act the hours an employee spends in training are working time unless four conditions are met together: the training
Are personality tests legal in hiring, and when does the ADA treat a personality test as a medical examination?
The ADA bars medical examinations of job applicants before a conditional offer, and a psychological test that can identify a mental disorder is
Is an employer-paid certification taxable to the employee, and when is it a tax-free working condition benefit?
A certification an employer pays for is taxable wages unless a specific rule excludes it. Two rules can exclude it. A job-related certification thatPilot until December 31, 2026
Register your company
Free for up to 5 people. A link to the company cabinet arrives by email within one business day.

