NYC Local Law 144: The Plain-English Guide for Small Employers
9 min read · Last reviewed 1 Aug 2026
General information, not legal advice. Laws in this area change; verify against the official sources at the end of this guide and confirm specifics with employment counsel.
New York City's Local Law 144 was the first law in the US to regulate AI hiring tools directly, and it remains the strictest. If you use any automated tool to screen candidates for a job based in NYC, the law expects three things from you: an annual independent bias audit of the tool, a public summary of that audit on your website, and advance notice to candidates. It has been enforced since July 2023, and there is no small-business exemption.
Most guides to LL144 are written for enterprise HR teams and their lawyers. This one is for the owner of a 5-to-50-person company who posted a job, got 200 applications, and started using software to get through the pile. Here is what the law actually covers, what each requirement means in practice, and the cheapest realistic path to meeting them. This is general information, not legal advice: confirm specifics with employment counsel.
Who Local Law 144 actually covers
The law applies to employers and employment agencies that use an automated employment decision tool, an AEDT, to screen candidates for hire or employees for promotion, for jobs located in New York City. An AEDT is any computational process derived from machine learning, statistical modeling, data analytics, or artificial intelligence that issues a simplified output (a score, classification, or recommendation) and is used to substantially assist or replace your discretionary decision making.
The phrase that matters is "substantially assist or replace." Under the city's rules, that covers relying solely on the tool's output, weighting the output more heavily than any other factor, or using it to overrule a human judgment. If a tool scores or ranks your applicants and you use that ranking to decide who moves forward, you should assume you are covered. A spell checker or a tool that only organizes resumes into folders is not an AEDT; a tool that scores candidates is.
Two things surprise small employers. First, there is no headcount threshold: a 4-person company using an AEDT for an NYC job has the same duties as a bank. Second, the job's location drives coverage, not yours. If the role is based in NYC, even partly, or is a remote role tied to an NYC office, the safest assumption is that the law applies. If you never hire for NYC-based roles, LL144 itself likely does not reach you, but Illinois, California, Colorado, and federal law have their own rules.
Requirement 1: the independent bias audit
Before using an AEDT, and at least annually after that, the tool must have had a bias audit within the prior year, done by an independent auditor. Independent means genuinely outside: not you, not the tool's vendor, and not anyone with a financial interest in the tool beyond the audit fee or who was involved in developing, distributing, or using it.
The audit itself is statistical. It computes selection rates for candidates by sex and by race/ethnicity (using the categories employers already know from EEO-1 reporting), plus intersectional combinations, and then the impact ratio: each group's rate divided by the rate of the most-selected group. For tools that output scores rather than binary decisions, the rules use a scoring-rate method based on how often each group scores above the median. The audit can use your historical data, historical data pooled across multiple employers that use the same tool, or test data if there is not enough history.
In practice, small employers rarely commission a solo audit. The realistic path is to ask your screening vendor for the latest independent audit of their tool and its distribution date, since the rules let you rely on an audit built on pooled historical data. If your vendor cannot produce one, that is a strong signal to switch tools. Whatever tool you use, you will make the auditor's job far easier if it produces a structured export of scores and outcomes per candidate. SiftFirst is built for that: every score carries quoted evidence, and the audit-ready export gives an auditor exactly the selection data they need to join with your demographic records.
Requirement 2: publish the audit summary
Having the audit is not enough: a summary of the most recent audit's results must be publicly posted on the employment section of your website before you use the tool, together with the distribution date of the tool. The summary needs to include the selection rates and impact ratios the audit computed and the date of the audit.
For a small business this is usually a single page or PDF linked from your careers page. If you rely on a vendor-level audit, the vendor typically provides a summary you can post or link. Keep it up for as long as you use the tool (and the rules expect it to stay available for a period after you stop). The point of this requirement is that a rejected candidate, or the city's enforcement agency, can check the audit exists without asking you.
Do not overthink the format, but do not skip it either: the audit-without-publication combination is one of the easiest violations for the Department of Consumer and Worker Protection to spot, because it is visible from the outside.
Requirement 3: the 10-business-day candidate notice
Candidates who live in New York City must be told, at least 10 business days before the AEDT is used on them, that an automated tool will be used, and which job qualifications and characteristics it will assess. The notice also has to include instructions for requesting an alternative selection process or a reasonable accommodation, and you must make information about the data the tool collects available on written request.
You can deliver the notice in the job posting itself, on your careers page, or by direct message to the candidate. The job-posting route is the simplest: put the notice in every posting for NYC roles and the 10-day clock takes care of itself for anyone who applies after it went up. Keep a record of what the notice said and when it was posted or sent.
The awkward part is listing the qualifications and characteristics the tool assesses. With a black-box matcher, you genuinely may not know. With a rubric-based tool you do: the criteria are the list. SiftFirst's free candidate notice generator builds the notice text directly from your screening criteria, which is exactly the disclosure the law is asking for.
Penalties and how enforcement actually happens
Violations carry civil penalties of $500 for a first violation and $500 to $1,500 for each subsequent one. The multipliers are what make it expensive: using a tool without a current audit counts as a separate violation each day, and failing to notify counts separately per candidate. A quiet month of screening a few hundred NYC applicants without notices can, on paper, add up to six figures.
Enforcement is run by the NYC Department of Consumer and Worker Protection and is largely complaint-driven. The people most likely to complain are rejected candidates, and AI screening produces a lot of rejected candidates. Separately from LL144, using a biased tool can create liability under the NYC Human Rights Law and Title VII, where damages are not capped at a fine schedule.
The honest summary: the direct penalties are survivable, but the paper trail you build for LL144 (audit, published summary, notices, records of who decided) is the same paper trail that protects you in a discrimination claim, which is the real financial risk.
What changed in 2026: the enforcement audit
For the first two years, the honest read on Local Law 144 was that the rules were strict and the enforcement was thin. That changed in December 2025, when the New York State Comptroller published an audit of how the Department of Consumer and Worker Protection had enforced the law between July 2023 and June 2025, and concluded that the enforcement was ineffective.
The specifics explain why the picture is about to look different. Auditors found that 75 percent of test calls made to 311 about automated hiring tools were misrouted and never reached the department at all, so complaints that should have started an investigation simply evaporated. They also re-examined the published bias audits of 32 companies and found at least 17 instances of potential non-compliance, including missing impact ratios, incomplete demographic breakdowns, and summaries that left out required information. The department's own review of those same companies had flagged one.
The department has accepted most of the audit's recommendations. So the practical position for an employer is the reverse of what it was: the reason many companies got away with a missing or incomplete audit summary was a complaint pipeline that did not work and a review process that did not catch much, and both are being fixed. If you concluded at some point that this law was not really enforced, that conclusion has an expiry date on it.
What this does not change is the substance. The duties are the same three they always were, and the cheapest path through them is unchanged: get the audit from your vendor, publish the summary with its distribution date, and send the notice. What has changed is the odds of anyone checking.
A practical compliance path for a small team
Here is the minimal sequence that gets a small NYC employer covered without a compliance department. First, inventory: list every tool that scores, ranks, or filters candidates, including features inside your job board or ATS you may have turned on without thinking of them as AI. Second, for each tool, get the vendor's latest independent bias audit and distribution date; post the summary on your careers page. Third, add the candidate notice to every NYC job posting, listing the criteria the tool assesses. Fourth, keep records: the rubric, each candidate's scores, and who made the final call.
Fifth, prefer tools that make all of this easy. A screening tool that shows its criteria, quotes evidence for every score, and never auto-rejects turns each requirement from a research project into an export. That is the design philosophy behind SiftFirst: you set the criteria, every score carries a quote from the resume, a human makes every decision, and the audit-ready export plus the notice generator produce the LL144 artifacts from data you already have.
Before you rely on any of this, run your specific setup past employment counsel, especially the questions of whether your tool "substantially assists" decisions and which of your roles count as NYC jobs. The two-minute compliance check on this site will show you which requirements likely apply to you and where your gaps are.
Key takeaways
- ✓LL144 applies to any employer using an automated tool to screen candidates for NYC-based jobs. There is no small-business exemption.
- ✓Three duties: an independent bias audit within the last year, a published summary of it on your careers page, and a candidate notice 10 business days before use.
- ✓You can usually rely on your vendor's pooled independent audit rather than commissioning your own; ask for it and post the summary.
- ✓Put the candidate notice in every NYC job posting: the tool in use, the qualifications it assesses, and how to request an alternative process.
- ✓Penalties run $500 to $1,500 per violation and accrue per day and per candidate; the bigger exposure is a discrimination claim, which the same records defend against.
- ✓A December 2025 State Comptroller audit found NYC enforcement ineffective: 75% of 311 complaints misrouted, and 17 potential violations where the department had flagged one. Most fixes were accepted, so assume checking gets more likely, not less.
Screening built for these rules
SiftFirst scores candidates against criteria you set, quotes the resume line behind every score, never auto-rejects, and exports the records these laws expect. The candidate notice generator and bias audit self-check are free.
FAQ
Does Local Law 144 apply if my company is outside NYC but I hire a remote candidate who lives there?
Coverage turns on where the job is located, not where your company is. A role based in an NYC office, even part-time, is covered, and remote roles tied to an NYC location can be. A fully remote role with no NYC connection is a grayer area, but candidate-notice duties attach to NYC residents. If you regularly get NYC applicants, the cheap move is to comply for everyone rather than litigate the boundary. Confirm your specific setup with counsel.
Is SiftFirst's bias audit self-check the official LL144 audit?
No. The official audit must be performed by an independent third-party auditor. SiftFirst's free self-check computes the same selection-rate and impact-ratio math in your browser so you can spot problems early and walk into the official audit prepared, but it does not satisfy the legal requirement by itself.
What if I only use AI to write job descriptions or interview questions?
LL144 targets tools that substantially assist or replace decisions about candidates: scoring, ranking, filtering, recommending. A tool that drafts your job description or suggests interview questions is not making an employment decision about a person, so it is generally outside the AEDT definition. The moment a tool starts ordering or filtering your applicant pile, assume it is in scope.