Does the law cover remote roles?

Yes, if the role is based in New York City. The trigger is the location of the role, not the candidate's location. A remote candidate applying for a position that reports to a NYC office is covered. A fully remote role with no NYC nexus is outside the law's scope. Seek legal guidance for borderline cases.

Can the AI vendor conduct the bias audit itself?

No. The law requires an independent auditor, and independence means the auditor cannot be the vendor or anyone with a financial interest in the tool's continued use. Many vendors offer their own internal bias testing reports. Those reports may be useful supplemental documentation, but they do not satisfy the annual independent audit requirement under Local Law 144.

What does the bias audit actually measure?

The audit must calculate the impact ratio across sex, race/ethnicity, and intersectional categories (e.g., Asian women, Black men) using EEOC definitions. The impact ratio is a group's selection rate divided by the highest-performing group's rate. Results must be published publicly and include the audit date, number of individuals assessed, and scoring methodology.

What are the penalties for non-compliance?

A first violation carries a $500 civil penalty. Each subsequent violation carries $500 to $1,500. Each day an AEDT is used without a valid audit is a separate violation, and each missed candidate notice is a separate violation, so penalties compound quickly across a hiring cycle. Employment agencies placing candidates in NYC roles are covered alongside direct employers.

Are employment agencies covered as well as direct employers?

Yes. The law covers both employers and employment agencies that use AEDTs in employment decisions for NYC-based positions. If you are a staffing agency screening candidates for NYC client roles using an AI tool, you are required to comply with the audit, publication, and notice requirements.

NYC Local Law 144 AI Bias Audit: Employer G…

Q: What is an automated employment decision tool (AEDT) under NYC Local Law 144?

Any computational process from machine learning, statistical modeling, or AI that issues a simplified output (score, classification, or recommendation) used to substantially assist or replace discretionary employment decisions. A resume ranking tool that scores applicants for human review qualifies. A keyword search that humans review independently generally does not.

NYC Local Law 144 AI Bias Audit: Employer Guide 2026

NYC Local Law 144 requires employers using AI hiring tools to conduct annual independent bias audits, publish results, and notify candidates. DCWP enforcement began July 5, 2023. Penalties run $500 to $1,500 per violation, with each day of use and each missed notice counted separately. Six-step compliance checklist.

TL;DR: NYC Local Law 144 requires employers using AI hiring tools to run annual independent bias audits, post results publicly, and notify candidates 10 business days before assessment. Penalties compound per applicant per day. Vendor testing does not count.

If your company uses software that scores, ranks, or filters job candidates in New York City, you are likely already subject to NYC Local Law 144, the first municipal law in the United States to require annual independent bias audits of AI hiring tools. The law's final rules took effect and DCWP enforcement began on July 5, 2023, and enforcement has been active since.

This guide explains what the law covers, who it applies to, what the audit must include, and the six steps you need to complete before your next hiring cycle.

What NYC Local Law 144 actually requires

Local Law 144 imposes three obligations on covered employers and employment agencies:

Annual independent bias audit. Any automated employment decision tool (AEDT) used in NYC hiring or promotion decisions must be audited by an independent third party at least once per calendar year. The audit must analyze whether the tool produces disparate results across sex and race/ethnicity categories.

Public publication of results. Audit results must be posted on the employer's public website and must remain accessible for at least six months after a new audit is completed. Results must include the date of the audit, the scoring methodology, the number of individuals assessed, and the calculated impact ratios.

Advance notice to candidates and employees. Anyone being assessed by an AEDT must receive written notice at least 10 business days before the assessment occurs. The notice must identify the tool being used, explain what data it collects and analyzes, and provide a way for the individual to request an accommodation or alternative process.

All three obligations apply simultaneously. An employer who completes the bias audit but skips the notice requirement is still non-compliant.

Who is covered: employers and employment agencies

The law covers any employer or employment agency that uses an AEDT to screen, rank, or evaluate candidates or employees for positions in New York City. Both categories are explicitly included, which means staffing firms and recruiting agencies placing candidates in NYC roles must also comply.

The geographic trigger is the location of the role, not the location of the hiring company or the candidate. If the position is based in New York City, or if the employee in question is employed in a NYC location, the law applies to the AEDT used in that decision.

Common misconception: some employers assume that because their headquarters are outside New York City, or because a role is labeled "remote," they are exempt. The relevant question is whether the role has a NYC location nexus. A remote role that reports to a NYC manager and could require occasional presence in a NYC office is likely covered. Legal counsel is advisable if you are unsure.

What counts as an AEDT

The law defines an AEDT as any computational process derived from machine learning, statistical modeling, data analytics, or AI that issues a simplified output, including a score, classification, or recommendation, that is used to substantially assist or replace discretionary employment decision-making.

In practice, this captures:

Resume screening tools that score or rank candidates
Video interview analysis platforms that assess facial expressions, tone, or word choice
Automated personality or cognitive assessments
Any tool that produces a pass/fail, score, or shortlist that hiring managers regularly act on

It does not capture tools that purely assist human judgment without producing a simplified output, such as a spell-checker or a calendar scheduling tool. The line turns on whether the tool substantially assists the decision. If hiring managers routinely follow the tool's rankings without independent review, the tool almost certainly qualifies as an AEDT.

Office interview setting representing the hiring process covered under NYC Local Law 144

What the bias audit must cover

The audit must calculate the impact ratio for each sex category, each race/ethnicity category, and each intersectional category (sex combined with race/ethnicity, such as Hispanic women or Black men) defined by the EEOC. The impact ratio is the selection rate of a given group divided by the selection rate of the highest-performing group. An impact ratio below 0.80, the so-called four-fifths rule from EEOC Uniform Guidelines, may indicate adverse impact, though the law does not prohibit any specific ratio level.

The auditor must be independent, meaning they cannot be the vendor of the tool, an affiliate of the vendor, or anyone with a financial interest in the tool's continued use. Many AEDT vendors provide internal bias testing reports, sometimes called fairness evaluations or algorithmic impact assessments. These are not substitutes for an independent audit. They may be useful background material for the auditor, but they do not satisfy the law's requirement.

Audit cost varies by tool complexity and auditor. Independent bias audits typically range from $5,000 to $50,000 per tool per year. Multi-tool environments require a separate audit for each covered AEDT.

The published audit results must be accessible to anyone without creating an account or paying a fee. A direct URL on your careers or legal compliance page is sufficient.

The candidate notice requirement in detail

The notice obligation is where many employers slip up. The 10-business-day window is a hard minimum. Notice must be delivered before the assessment occurs, not at the time of assessment or afterward.

Notice must include:

The fact that an AEDT is being used in the employment decision
The type of data it collects and uses
A description of the simplified output it produces
Instructions for how the candidate can request an accommodation or alternative selection process

For most employers, this means updating job application confirmation emails or screening intake pages to include the required information. It is also good practice to document when notice was sent and by what method, since enforcement may require you to demonstrate compliance.

Employment agencies must ensure they provide notice not just on behalf of their own process but also in relation to any AEDT used by a client employer if the agency is the party deploying the tool.

Penalties: how the math works

The civil penalty structure under Local Law 144 compounds quickly.

First violation, and each additional violation on the same day: $500
Each subsequent violation: $500 to $1,500

Two things make this compound. Each day an AEDT is used without a valid bias audit is a separate violation, and each failure to provide the required candidate notice is a separate violation. DCWP has discretion over exactly how violations are counted, but across a normal hiring cycle, with screening happening daily and notice owed to every candidate, the totals can climb into the tens or hundreds of thousands of dollars. Treat the per-violation structure, not any single headline figure, as the warning.

Enforcement is complaint-driven. The NYC Department of Consumer and Worker Protection (DCWP) handles enforcement and can impose these penalties following investigation. Getting the process right from the start is cheaper than remediation after a complaint.

Six-step compliance checklist

Use this checklist before your next hiring cycle that involves any AI or algorithmic screening tool.

Step 1: Inventory your AEDTs. List every software tool used in candidate screening or promotion decisions for NYC roles. Include resume screening platforms, video interview tools, assessments, and any tool that produces a score or ranking. If your ATS has an AI scoring module, that module is likely an AEDT even if the ATS as a whole is not.

Step 2: Confirm NYC nexus. For each tool in your inventory, identify which roles it is used for and confirm whether any of those roles are located in New York City. Document your analysis, especially for remote roles where the nexus may be arguable.

Step 3: Commission the bias audit. Identify an independent auditor with no financial relationship to your AEDT vendor. Provide the auditor with access to the tool's outputs and any available demographic data from your candidate pool. The audit must cover the preceding 12 months of use.

Step 4: Publish results. Once the audit is complete, post the required data on a publicly accessible page of your website. Include the audit date, methodology, number of individuals assessed, and impact ratios by sex and race/ethnicity category.

Step 5: Build the notice workflow. Update your candidate communication process to deliver the required notice at least 10 business days before any AEDT assessment occurs. Confirm the notice includes all required elements: tool description, data collected, output type, and accommodation request process.

Step 6: Document everything. Keep records of audit contracts, audit reports, publication dates, and notice delivery logs. If DCWP investigates a complaint, your documentation is your defense.

Common compliance mistakes

Assuming vendor testing counts. It does not. Independent means independent. If your vendor has excellent internal fairness documentation, that is good context for your auditor, but you still need to commission your own audit.

Missing remote role coverage. The remote-role exemption is narrower than most employers assume. When in doubt, treat the role as covered and document your reasoning if you decide it is not.

Skipping the notice requirement. Many employers complete the audit and publication steps but forget the candidate notice obligation. All three requirements must be met. An audit alone does not make you compliant.

Late notice. Ten business days is the minimum. If you send notice the day before an assessment, you are in violation regardless of the content of the notice.

Not re-auditing after tool changes. If your vendor updates the model or the tool's scoring methodology, the prior audit may no longer accurately represent the current tool. Best practice is to treat a significant model update as a trigger for a new audit cycle.

How this fits with other AI hiring laws

NYC Local Law 144 does not exist in isolation. Employers operating in multiple states face overlapping requirements. Illinois requires written notice and consent before AI analyzes video interviews. Connecticut's SB 5 (the CART Act, Public Act 26-15) brings an automated employment decision technology framework with principal AEDT obligations effective October 1, 2027, including a rule that use of such a tool is not a defense to a discrimination claim. (The October 1, 2026 provision covers only WARN Act AI disclosure for layoffs.) Colorado's ADMT framework takes effect January 2027.

Federal law also applies. Title VII, the ADA, and the ADEA all apply to AI hiring tools regardless of whether the discrimination was caused by an algorithm. The four-fifths rule from EEOC Uniform Guidelines applies to AI selection procedures just as it does to traditional assessments.

For a full state-by-state matrix, see the AI hiring tool compliance US state laws guide. For overlapping federal obligations, the EEOC AI hiring guidance 2026 employer checklist covers the federal layer.

If your AI screening vendor is involved in litigation, the Workday AI lawsuit HR screening checklist covers what that means for employers relying on third-party tools. For FCRA obligations that run alongside Local Law 144 for background screening contexts, see FCRA AI hiring disclosure requirements 2026.