NYC Local Law 144: What AI Hiring Compliance Actually Requires

NYC Local Law 144 requires any employer or employment agency using an automated tool to screen or rank candidates for NYC-based or NYC-associated roles to run an independent annual bias audit, publish a summary of the results, and give candidates 10 business days’ notice before use — or risk fines of $375 to $1,500 per violation.

Since July 5, 2023, New York City has required employers to audit the AI and algorithmic tools they use to screen job candidates, not just for effectiveness, but for discriminatory impact. Most enterprise hiring stacks now include at least one system that meets the law’s definition, whether that’s an applicant-tracking platform’s built-in resume scoring, a video-interview analysis tool, or a dedicated skills-assessment product. The compliance obligation sits with the employer, not the vendor — which means “our ATS vendor handles that” is not a defense DCWP recognizes.

What counts as an “Automated Employment Decision Tool” (AEDT)

The law’s definition is broader than most hiring teams assume. An AEDT is any computer-based tool that uses machine learning, statistical modeling, data analytics, or AI, and that substantially assists or replaces discretionary decision-making in hiring or promotion. Critically, the law applies “at any point in the process” — a resume-screening filter that narrows 500 applicants to 50 is in scope even if a human makes the final call, because it substantially assisted the earlier discretionary step. (NYCRules)

Tools that commonly qualify: resume-parsing and ranking software, automated video-interview scoring, chatbot-based candidate screening, skills or personality assessment platforms with algorithmic scoring, and any internal model used to rank or shortlist promotion candidates.

Who has to comply

The obligation applies to employers and employment agencies with a physical office in New York City that use an AEDT for either of these:

• A job located in NYC, full- or part-time.
• A remote position that is associated with a NYC office — a detail many out-of-state or hybrid employers miss, since remote does not automatically mean out of scope.

The law does not apply to outreach messages or generic invitations to apply — only to tools used in the actual screening, scoring, or ranking decision. (Gibsondunn)

The annual bias audit: what it has to measure

The bias audit is the law’s centerpiece, and it has specific technical requirements, not just a general fairness review:

• Independence. The auditor must exercise objective, impartial judgment — they cannot be an employee of the company or vendor, cannot have been involved in developing the AEDT, and cannot hold a direct or material indirect financial interest in either party. (NYC)
• What’s measured. The audit calculates selection or scoring rates and the resulting impact ratio across sex categories, race/ethnicity categories, and intersectional categories (e.g., Black women vs. white men), not just single-category comparisons.
• Data source. Audits must use historical usage data or, where unavailable, test data — employers may not infer or impute a candidate’s demographic information to manufacture audit data.
• Scope flexibility. Audits don’t need to be run separately for every open role; data can be aggregated across similar positions.
• Refresh cycle. An audit is only valid for one year from the date it was conducted. There is no grace period for “we’re between audits” — a tool used past the one-year mark without a fresh audit is out of compliance.

Notice requirements: the 10-business-day rule

Employers must notify NYC-based candidates and employees that an AEDT will be used, at least 10 business days before it’s used, via job posting, mail, email, or a website notice. The notice must also include instructions for requesting a reasonable accommodation under other applicable laws (e.g., if a candidate needs an alternative to an AI-scored assessment for a disability-related reason). Notices can describe the tool and job category in general terms — they don’t need to be rewritten for every individual req.

Public posting: what has to go on your website

Separately from candidate notice, employers must publish on their website (or via a clearly linked page) a summary of the results of the most recent bias audit and the date the AEDT was first used (“distribution date”). This is a standing, always-current obligation — not a one-time filing.

Compliance at a glance

Penalties — and why “rarely enforced” doesn’t mean “safe”

Violations of the audit, notice, or posting requirements carry fines ranging from $375 to $1,500 per violation, with each day of continued non-compliance typically treated as a separate violation. In practice, enforcement so far has been light: a December 2025 audit by the New York State Comptroller’s office found that NYC’s Department of Consumer and Worker Protection (DCWP) received only two AEDT-related complaints between July 2023 and June 2025, and had not investigated whether its own complaint-intake process was working correctly. (Littler)

The same audit is a better argument for compliance than enforcement history would suggest, though. When DCWP reviewed a sample of 32 companies’ websites and bias-audit postings, it flagged just one non-compliance issue. When the Comptroller’s own audit team reviewed the same 32 companies, they found at least 17 instances of potential non-compliance — more than half the sample. The gap wasn’t a legal loophole; it was that DCWP staff lacked the technical background to evaluate AEDTs correctly and hadn’t consulted the city’s own tech agency for help. The auditors recommended DCWP build proactive review processes and stop relying almost entirely on complaints it rarely receives. (OSC)

The practical takeaway: low complaint volume reflects weak detection, not widespread compliance. As enforcement capacity catches up — which the audit explicitly pushes DCWP to do — the exposure for companies that assumed “nobody’s checking” doesn’t disappear; it becomes retroactive.

A practical compliance checklist

1. Inventory every tool that touches a hiring or promotion decision — not just your ATS, but video-interview platforms, skills assessments, chatbots, and any internal scoring model — and determine which meet the AEDT definition.
2. Confirm scope: does the role sit in NYC, or is a remote role associated with a NYC office? Don’t assume remote means exempt.
3. Commission an independent bias audit from an auditor with no financial or developmental tie to the tool or your company, calculating impact ratios across sex, race/ethnicity, and intersectional categories.
4. Calendar the one-year audit expiration the day the audit is completed — not the day you remember to check.
5. Draft and send the 10-business-day notice to affected candidates/employees, including accommodation-request instructions, before the AEDT goes live for a given hiring cycle.
6. Publish the audit summary and distribution date on your website, and update it every time a new audit completes.
7. Log all of the above — audit reports, notice records, posting update history — so you can demonstrate compliance if DCWP’s enforcement posture changes, per the Comptroller’s recommendations.

Next step: See how Fruggr helps map your AI hiring tools to Local Law 144 obligations — or request a demo to see the full AI governance platform.

FAQ

This article was drafted with AI assistance and reviewed by our editorial team before publishing.