NYC Local Law 144: Automated Employment Decision Tools
Requires an annual independent bias audit of hiring/promotion AI tools, publication of results, and advance notice to candidates. It is the most-cited US example of mandatory algorithmic bias auditing.
Overview
The duty is triggered by a narrow technical test rather than by any use of AI: DCWP reads "substantially assist or replace discretionary decision making" as relying solely on a simplified output, weighting that output more heavily than any other criterion, or using it to overrule conclusions reached by other factors including human judgement. Audit content is prescribed rather than left to the auditor, which is unusual for a transparency law: the audit must report selection or scoring rates and impact ratios across the EEO-1 component 1 groups, broken out by sex, by race and ethnicity, and by intersectional combinations of the two. Data provenance is also fixed, with historical data from real use of the tool as the default and synthetic test data permitted only where there is too little history for a statistically significant result. Enforcement is complaint driven, handled by the Department of Consumer and Worker Protection, and heard before the Office of Administrative Trials and Hearings, with each day of non-compliant use counting as a fresh violation and each missed notice counting separately. Passing an audit buys no immunity: the subchapter expressly preserves private civil actions and the jurisdiction of the NYC Commission on Human Rights.
Key dates
- Dec 11, 2021Enacted
- Jul 5, 2023Enforcement began
Risk areas addressed
Who it applies to
Any employer or employment agency using an automated employment decision tool to screen candidates for employment, or employees for promotion, within New York City is covered, with no headcount, turnover or sector threshold; the notice duties attach specifically to candidates and employees who reside in the city. Tool vendors carry no direct liability under the subchapter, although in practice they commission the audit that their customers rely on, and a single audit can cover multiple employers where each contributes its own historical data. Out of scope are tools that neither automate nor substantially assist discretionary decisions, such as spam filters, firewalls, calculators, spreadsheets and plain data sets, along with outputs from tools that only translate or transcribe, for example converting a CV from PDF or transcribing a recorded interview.
Key obligations
- Obtain a bias audit of each tool from an independent auditor no more than one year before use, and stop using the tool once more than a year has elapsed since the most recent audit (6 RCNY § 5-301(a)).
- Confirm auditor independence before engagement: an auditor is disqualified if they were involved in using, developing or distributing the tool, hold an employment relationship with you or the vendor during the audit, or hold a direct or material indirect financial interest in either (6 RCNY § 5-300).
- Require the audit to calculate selection or scoring rates and impact ratios for every category, computed separately for sex, for race and ethnicity, and for intersectional categories of sex, ethnicity and race, and to state the number of assessed individuals excluded because they fall into an unknown category (6 RCNY § 5-301(b)).
- Base the audit on historical data from actual use of the tool, and use test data only where there is insufficient historical data for a statistically significant audit, in which case the summary must explain why and describe how the test data was generated and obtained (6 RCNY § 5-302).
- Publish, on the employment section of your website before the tool is used, the date of the most recent audit and a summary of results covering the source and explanation of the data, the number of applicants or candidates, the selection or scoring rates and the impact ratios for all categories, plus the tool's distribution date (6 RCNY § 5-303(a)).
- Keep that summary and distribution date posted for at least six months after the last use of the tool for an employment decision (6 RCNY § 5-303(c)).
- Notify each candidate or employee resident in the city at least 10 business days before use, covering both the fact that a tool will be used and the job qualifications and characteristics it will assess, and include instructions for requesting an alternative selection process or a reasonable accommodation (§ 20-871(b)(1) and (2); 6 RCNY § 5-304(a)).
- Post your data retention policy, the type of data collected for the tool and the source of that data on the employment section of your website, with instructions for making a written request, and answer any written request within 30 days (§ 20-871(b)(3); 6 RCNY § 5-304(d)).
How to prepare
- Inventory every tool that scores, ranks, tags or classifies applicants or promotion candidates for New York City roles, then apply the three-part DCWP test to each so you can document which are in scope and which are not.
- Engage an auditor, record the independence checks in writing, and hand over the prior 12 months of your own selection data broken down by EEO-1 category so the audit runs on historical data rather than synthetic test data.
- Commission the audit to produce selection or scoring rates and impact ratios at sex, race and ethnicity, and intersectional level, and require written justification plus applicant counts for any category the auditor excludes as representing under 2% of the data.
- Publish the results summary and distribution date on the employment section of the careers site before the tool goes live, or point to it with a hyperlink clearly labelled as the bias audit results.
- Insert the 10 business day notice, the qualifications and characteristics disclosure and the alternative process instructions into job postings or candidate emails, and place the data type, data source and retention information on the same web page.
- Diarise the next audit for 11 months out, and keep the published summary live for six months after you retire the tool.
Penalties & enforcement
Civil penalties per violation, with a first violation and any further violation on the same day capped at $500, rising to $1,500 for subsequent violations. Each day of non-compliant use and each missed notice counts separately (§ 20-872).
Enforced by: NYC Department of Consumer and Worker Protection
NYC LL 144 (Bias Audit) is enforced by NYC Department of Consumer and Worker Protection. Its obligations are already live, so a gap is a present exposure rather than a future one. In practice that means knowing which of your AI systems fall in scope in employment & HR, holding assessments that speak to bias & discrimination, automated decisions and transparency, and being able to produce that evidence on request.