NYC Local Law 144 Enforcement: AI Hiring Audits Required

NYC Local Law 144 First Enforcement Actions: AI Hiring Bias Audits Are No Longer Optional

New York City Local Law 144—the first law in the United States to mandate bias audits for artificial intelligence tools used in employment decisions—entered enforcement mode in 2023, and regulators have begun issuing violations for non-compliance. The first enforcement actions confirmed what employment law attorneys had been warning: the era of voluntary AI hiring compliance is over. If you use any automated employment decision tool in New York City, bias audits are now legally mandatory.

According to the New York City Department of Consumer and Worker Protection (DCWP), Local Law 144 covers any "automated employment decision tool" (AEDT) used to substantially assist or replace discretionary decision-making in hiring. Employers who use covered tools without completing required bias audits face penalties of $375–$1,500 per violation per day.

This article analyzes the first LL 144 enforcement actions, what they reveal about DCWP's enforcement priorities, and the compliance steps every New York City employer must take before using AI in hiring. For a running list of AI hiring enforcement across the U.S., see our AI hiring lawsuits tracker.

What Is NYC Local Law 144?

NYC Local Law 144, codified at NYC Admin. Code § 20-871 et seq., took effect on July 5, 2023, following a series of extensions from the original January 2023 effective date. The official compliance framework is published by the NYC Commission on Human Rights (CCHR). The law:

1. Defines automated employment decision tools (AEDTs) as computational processes derived from machine learning, statistical modeling, data analytics, or artificial intelligence that is used to substantially assist or replace discretionary decision-making in screening candidates for employment or employees for promotion
2. Requires employers to conduct independent bias audits of AEDTs before use and annually thereafter
3. Mandates public disclosure of bias audit results on employer websites
4. Requires notice to candidates and employees that AEDTs are being used
5. Prohibits use of non-audited AEDTs in NYC employment decisions

Key Definitions

Automated Employment Decision Tool (AEDT): A machine learning, statistical modeling, data analytics, or AI-based process that substantially assists or replaces human decision-making in hiring or promotion. This includes:

• Resume screening software with algorithmic ranking
• Automated video interview analysis
• Candidate scoring platforms
• Skills assessment tools with algorithmic scoring
• Personality or cognitive assessment tools with algorithmic outputs

Bias Audit: An "impartial evaluation" by an "independent auditor" of an AEDT's performance across categories including sex, race/ethnicity, and intersectional categories. The audit must calculate selection rates and impact ratios for each covered category.

Impact Ratio: The selection rate of a category divided by the selection rate of the most-selected category. An impact ratio below 0.8 (the "four-fifths rule") suggests adverse impact.

The First Enforcement Actions: What DCWP Found

When DCWP began enforcement in mid-2023, investigators found widespread non-compliance. The enforcement pattern revealed several common violations:

Violation 1: Using AEDTs Without Any Audit

Many employers had never considered whether their hiring software constituted an AEDT subject to LL 144. Resume parsing tools, applicant tracking systems with automated scoring, and video interview platforms all potentially qualify—and many employers using these tools had conducted no bias audits.

Violation 2: Conducting Insufficient Audits

Some employers had hired consultants to conduct "bias assessments" that did not meet LL 144's specific requirements. DCWP enforcement established that a qualifying audit must:

• Be conducted by an "independent" auditor (no conflict of interest)
• Calculate selection rates by sex and race/ethnicity
• Calculate impact ratios for each category
• Calculate intersectional rates (e.g., Black women separately from Black men and white women)
• Cover the specific tool being used, not just AI systems generally

Violation 3: Failing to Publish Results

Even employers who conducted qualifying audits often failed to publish results on their websites in a "clear and conspicuous" manner. DCWP requires the audit results to be:

• Published on the employer's publicly accessible website (or the AEDT vendor's website)
• Maintained until at least 6 months after the employer stops using the AEDT
• Including the date of the most recent bias audit and the summary of results

Violation 4: Inadequate Candidate Notice

LL 144 requires employers to notify candidates that an AEDT will be used no less than 10 business days before the tool is used. The notice must:

• State that an AEDT is being used
• Describe the data collected and how it is used
• Explain the candidate's right to request an alternative assessment process

Many employers were providing this notice too late or with insufficient detail.

Penalties: What Non-Compliance Costs

LL 144 establishes a tiered penalty structure:

For a large employer using a non-audited AEDT in NYC hiring for 30 days across 5 locations, the penalty exposure is: $375 (day 1) + ($1,500 × 29 days × 5 locations) = $218,875.

This is not theoretical—DCWP has demonstrated willingness to pursue daily violations for sustained non-compliance.

Who Is Covered: AEDT Scope Under LL 144

Covered Employers

LL 144 applies to employers who:

• Use AEDTs to screen candidates for employment in New York City
• Use AEDTs to screen current employees for promotion in New York City

The law covers employers regardless of size and regardless of whether they are incorporated in New York City.

Covered Tools

The "substantially assists or replaces discretionary decision-making" standard has been interpreted broadly by DCWP. Tools that may be covered include:

• Resume screening algorithms (HireVue, Workday's algorithmic scoring, Greenhouse's automated ranking)
• Video interview AI (HireVue, Spark Hire with AI scoring features)
• Cognitive and personality assessments with algorithmic outputs (Pymetrics, Korn Ferry, Criteria Corp)
• Chatbot screening tools that advance or eliminate candidates based on responses
• ATS scoring features that rank candidates without human review of individual rankings

Tools that are likely not covered:

• Traditional (non-AI) skills tests with manual scoring
• Human-reviewed assessments where the assessor has full discretion
• Search tools that surface candidates without ranking or scoring them

Gray Areas

DCWP's guidance acknowledges significant ambiguity about which tools qualify. When in doubt, assume coverage and conduct a bias audit. The cost of an unnecessary audit ($5,000–$25,000) is far less than the cost of enforcement action ($375–$1,500/day per location).

How to Conduct a Compliant LL 144 Bias Audit

Step 1: Identify Your AEDTs

Inventory every software tool used in your NYC hiring and promotion processes using our AI hiring compliance checklist:

• Document what each tool does
• Identify which tools use machine learning, statistical modeling, or AI
• Determine whether each tool substantially assists or replaces human decision-making
• Flag tools that likely qualify as AEDTs

Step 2: Select an Independent Auditor

The auditor must be:

• Independent (no financial relationship with the AEDT developer or employer that would compromise impartiality)
• Technically capable of analyzing the tool's selection rate data
• Available to provide a written audit report

Many AEDT vendors offer "vendor audits" through affiliated providers. These typically do not meet LL 144's independence requirement. Use a genuinely independent third-party auditor.

Step 3: Gather the Required Data

The auditor needs historical outcome data showing:

• How many candidates were assessed by the tool
• Which candidates were selected (advanced) or not selected
• Demographic data for assessed candidates (sex and race/ethnicity at minimum)

Data can come from two sources:

• Historical data: Prior assessments with actual outcome data (preferred)
• Test data: A representative test population if historical data is unavailable or insufficient

Step 4: Calculate Selection Rates and Impact Ratios

The audit must calculate:

• Selection rate for each sex category (male, female, non-binary if available)
• Selection rate for each race/ethnicity category (as defined in the NYC EEO forms)
• Selection rate for intersectional categories (e.g., Hispanic female separately from Hispanic male and white female)
• Impact ratio for each category (selection rate ÷ most-selected category's selection rate)

An impact ratio below 0.8 for any category indicates potential adverse impact requiring further analysis and remediation.

Step 5: Publish Results

Post the bias audit results on your company website. Required elements per the NYC CCHR:

• Date of the most recent bias audit
• Summary of results (selection rates and impact ratios)
• The date the AEDT was first used

Step 6: Provide Candidate Notice

At least 10 business days before using the AEDT on any candidate:

• Notify the candidate an AEDT will be used
• Describe the type of data collected
• Describe how data will be used
• Describe the candidates' right to request alternative selection processes

Step 7: Maintain Records

Retain bias audit documentation and candidate notice records for at least 6 months after the employer stops using the AEDT.

Comparison: AI Hiring Audit Requirements by Jurisdiction

For the full picture, see AI hiring laws by state:

New York City is the most demanding jurisdiction in the United States for AI hiring tool compliance. If you hire in NYC, LL 144 compliance is mandatory—and compliance processes you build for NYC will position you well for compliance with emerging state laws.

What Major AEDT Vendors Are Doing

Following LL 144's implementation, major AEDT vendors have responded differently:

HireVue: Published bias audit results on its website and updated client contracts to assist with LL 144 compliance. Clients still must ensure the audit data covers their specific implementation.

Workday: Updated its Recruiting product to include diversity analytics and provide data needed for client-conducted bias audits. The audit itself remains the employer's responsibility.

Pymetrics (now Harver): The vendor conducts regular bias audits and publishes aggregate results, but individual employer implementations may show different patterns requiring employer-specific audits.

Many smaller vendors: Have not conducted qualifying audits and are relying on clients to conduct their own. This is a compliance risk for employers—the obligation to audit runs to the employer, not just the vendor.

The EEOC has reinforced that employer liability is non-delegable: you cannot satisfy your obligations by pointing to a vendor's audit. And the OFCCP reviews AI tool usage for federal contractors during routine compliance audits, separate from DCWP enforcement.

The "Alternative Process" Requirement: What It Means in Practice

LL 144 requires employers to offer candidates and employees an alternative to the AEDT-based process upon request. Per NYC CCHR guidance:

• The alternative process must be a genuinely different pathway to evaluation—not just a re-run of the same algorithmic tool
• Employers must notify candidates of their right to request the alternative
• Employers must respond to requests within a reasonable timeframe
• The alternative process must be documented

In practice, this means most employers need both an AI-assisted screening track and a traditional human-review track for NYC candidates. This adds operational complexity but is legally required.

What the First Enforcement Actions Tell Us About DCWP's Priorities

Based on the initial enforcement pattern, DCWP is prioritizing:

1. Employers who knew about LL 144 but didn't act. Larger employers, particularly in technology, finance, and professional services, face heightened scrutiny because they are more likely to be using sophisticated AEDTs and are less likely to be able to claim unawareness.

2. Non-disclosure violations. Failure to publish bias audit results publicly is an easy-to-identify violation. DCWP investigators can check employer websites without any investigation, making this a common enforcement target.

3. High-volume hiring contexts. Employers conducting large-scale hiring in NYC—retail, logistics, customer service, tech—face higher exposure because the per-day, per-location penalty structure amplifies quickly.

4. Repeat violations. The penalty jumps from $375 to $1,500/day for subsequent violations. Employers who receive a violation notice and continue non-compliant practices face rapidly escalating penalties.

Frequently Asked Questions

When did NYC Local Law 144 take effect?

Local Law 144 took effect on July 5, 2023, following multiple extensions from the original January 2023 effective date. DCWP began enforcement actions in 2023, with violations issued for conduct occurring after July 5, 2023. Employers who have been using covered AEDTs since July 2023 without completing bias audits have been in violation since that date—and penalties are calculated from the date of non-compliance.

Does LL 144 apply to remote workers in NYC?

Yes. LL 144 applies to employment decisions affecting candidates and employees who would work in New York City, regardless of where the employer is headquartered or where the hiring decision-maker is located. If you are hiring a remote employee who will work in NYC, LL 144 applies. If you are hiring a fully remote employee who will never work in NYC, LL 144 likely does not apply—but this analysis should be confirmed by counsel.

Do we need to audit our ATS if we use it in NYC?

It depends on how your ATS is configured and used. If your ATS includes automated scoring, ranking, or filtering features that substantially assist or replace human decision-making in screening candidates, it likely qualifies as an AEDT requiring a bias audit. If your ATS is used purely as a database for storing and retrieving applications, with all screening done by human reviewers, it likely does not qualify. Most modern ATS platforms fall somewhere in between—review your specific implementation with counsel. See our compliance FAQ for more guidance.

Our vendor says they've already conducted a bias audit. Are we covered?

Probably not. LL 144 places the compliance obligation on the employer, not the vendor. A vendor-conducted bias audit may not satisfy the "independent auditor" requirement if the auditor has a financial relationship with the vendor. Additionally, aggregate vendor audit data may not reflect how the tool performs in your specific implementation with your candidate pool. Most employers need to conduct their own audit—though vendor audit data can be a useful input to your auditor.

What happens if our bias audit reveals adverse impact?

Discovering adverse impact is not itself a violation—it's the point of the audit. What matters is what you do with the information. If a bias audit reveals that your AEDT produces adverse impact against a protected class, you must take remedial action: recalibrate the tool, supplement with alternative assessment methods, or discontinue use. Document your remediation steps. Continued use of a tool with known adverse impact after completing an audit that identified the problem could be treated as intentional discrimination under Title VII (42 U.S.C. § 2000e).

Is there a minimum sample size for a valid bias audit?

DCWP guidance does not specify a minimum sample size, but the audit must use data sufficient to produce statistically meaningful results. Small employers with limited historical data may need to use test data or aggregate data across multiple employers. Work with your independent auditor to determine whether your available data is sufficient for a valid audit.

How much does a LL 144 bias audit cost?

Costs vary significantly based on the complexity of the tool, the volume of data, and the scope of the audit. Simple audits for single tools with readily available historical data may cost $3,000–$8,000. Complex audits covering multiple tools or requiring data collection and analysis infrastructure may cost $15,000–$50,000. Annual re-audits are typically less expensive than initial audits. These costs must be weighed against penalty exposure of $375–$1,500 per day per location—for most employers, compliance is significantly less expensive than enforcement.

Key Takeaways

• NYC Local Law 144 is in enforcement mode as of July 2023. Employers using covered AEDTs without completing bias audits face daily penalties of $375–$1,500 per location. Review official requirements at the NYC CCHR.

• "AEDT" is broadly defined. Resume screening software, video interview AI, and algorithmic candidate scoring tools all potentially qualify. When in doubt, treat the tool as covered and conduct an audit.

• The compliance obligation runs to the employer, not the vendor. Vendor-conducted audits may not satisfy the independence requirement; employers typically need to commission their own audits.

• Adverse impact discovery requires action, not just disclosure. Finding adverse impact in a bias audit triggers an obligation to remediate or discontinue the tool.

• Alternative assessment processes are required. Candidates have the right to opt out of AEDT-based screening; employers must have a genuine alternative pathway available.

• NYC is the strictest U.S. jurisdiction, but others are following. The Colorado AI Act (2026), Illinois AI hiring law, and California's proposed AB 2930 have similar requirements taking shape. Review AI hiring laws by state and your state employment compliance obligations to get ahead of what's coming.

EmployArmor helps NYC employers identify covered AEDTs, coordinate independent bias audits, publish compliant results pages, and maintain the candidate notice process required by Local Law 144. Get your free compliance assessment →

Last updated: March 2026. This content is for informational purposes only and does not constitute legal advice. Consult qualified employment counsel for guidance specific to your NYC compliance obligations.