Introduction to NYC Local Law 144 and AI Hiring Compliance

New York City Local Law 144, which took effect on July 5, 2023, represents one of the most significant regulatory frameworks governing the use of artificial intelligence in employment decisions. The law specifically targets the use of automated employment decision tools (AEDTs) by employers and employment agencies operating within the five boroughs. Its primary objective is to mitigate algorithmic bias and ensure that AI-driven hiring processes do not discriminate against protected classes based on race, gender, age, or other characteristics covered under the New York City Human Rights Law. The legislation mandates that before an AEDT is used to screen candidates or make employment decisions, the tool must undergo a bias audit. This audit must be conducted no more than one year prior to the tool's deployment, and the results must be made publicly available on the employer's website. The law applies to any entity that uses computational processing, including machine learning, statistical modeling, or AI, to make decisions about hiring, promotion, or termination. Non-compliance carries significant financial penalties, with fines reaching up to $1,500 per violation per day. As of September 2026, the law remains a cornerstone of NYC's approach to algorithmic accountability, forcing technology providers and HR departments to rigorously vet their AI systems. The audit process is not merely a bureaucratic checkbox; it is a substantive analysis of how an algorithm's outcomes correlate with demographic data, requiring deep technical understanding and legal foresight.

Also worth reading: What is an algorithmic adverse impact compliance checklist and how can employers use it to meet AI hiring regulations? · What is the definitive AI recruitment audit checklist for 2026 to ensure labor law compliance? · What does a religious organization employment compliance checklist look like in 2026, and how do AI tools streamline it?

The Audit Requirement: Scope and Timing

The specific audit requirement under Local Law 144 is deceptively simple in its wording but complex in its execution. The law stipulates that an independent auditor must conduct a bias audit of the AEDT no more than 12 months before the tool is used. This means that if an employer implements a new AI screening tool in 2026, the audit must have been performed at some point between July 2025 and July 2026. The audit must evaluate the tool's impact on protected groups, specifically looking at the selection rate for applicants within each group. The core metric calculated is the "impact ratio," which compares the selection rate of the most favored group to the selection rate of each other group. If the impact ratio falls below 80 percent, the law presumes disparate impact, triggering a requirement for the employer to demonstrate that the tool is job-related and consistent with business necessity. This 80 percent threshold, often referred to as the "four-fifths rule" in employment law, is a critical benchmark that auditors and employers must monitor closely. Furthermore, the audit must be performed by an independent auditor who is not the employer, not the developer of the tool, and does not have a financial interest in the AEDT's success. This independence requirement is designed to ensure objectivity and prevent conflicts of interest from compromising the audit's validity. The auditor must then issue a report detailing the methodology, the data used, and the conclusions reached regarding bias. This report must be made publicly available on the employer's careers page or website, ensuring transparency to current and prospective employees. The timing and independence requirements are the most litigated aspects of the law, as errors in these areas can render the entire compliance effort meaningless in the eyes of the NYC Department of Consumer and Worker Protection (DCWP).

Step-by-Step: Assembling the Audit Checklist

Creating a compliant audit checklist for NYC Local Law 144 requires a systematic approach that bridges the gap between data science and legal compliance. The first step involves identifying all AEDTs in use or planned for use. This encompasses not only obvious tools like resume screening software and video interview analysis platforms but also less apparent systems that use AI to rank candidates, predict job performance, or automate scheduling decisions. Once the tools are identified, the employer must gather the necessary data for the audit. This typically includes historical hiring data, the criteria used by the algorithm, and outcome data showing which candidates were selected or rejected. The auditor will then analyze this data to calculate selection rates for different demographic groups. A critical component of this phase is data quality; if the historical data used to train or validate the AI is itself biased, the audit will likely reveal disparities that could lead to legal liability. The auditor must also verify that the AEDT's predictions are consistent across different subgroups. For instance, if an AI tool screens out 30 percent of male applicants but 50 percent of female applicants for the same role, the impact ratio would be calculated, and if it falls below 80 percent, the tool is presumed biased. The checklist must therefore include a verification step where the employer confirms that they have reviewed the raw data for anomalies, missing values, or preprocessing errors that could skew the results. Finally, the auditor compiles a comprehensive report that not only states whether the tool passes or fails the 80 percent threshold but also provides context, limitations, and recommendations for mitigation if bias is detected. This report is the deliverable that must be posted publicly, making the integrity of the audit process paramount.

Comparison of Audit Methodologies: Independent vs. Internal

When organizations approach the Local Law 144 audit, they face a fundamental choice: conduct the audit internally using their own staff or hire an independent third-party auditor. Each approach carries distinct advantages, risks, and compliance implications that must be weighed carefully. An internal audit might seem cost-effective initially, as it leverages existing staff who are familiar with the company's operations and the specific AI tool in question. However, the law explicitly requires that the auditor be independent, meaning they cannot be an employee of the company utilizing the AEDT. Attempting to label an internal HR or IT staff member as an "independent auditor" would likely be scrutinized heavily by the DCWP during an investigation or audit. The risk of perceived bias is high, especially if the internal team was involved in selecting or implementing the AI tool. On the other hand, engaging an independent auditor provides a layer of legal protection and credibility. An external firm specializing in employment law and AI compliance brings a standardized methodology and an objective perspective that internal staff may lack. They are also more likely to stay current with evolving legal interpretations and technical standards for bias testing. The comparison between these two paths often comes down to the trade-off between cost and compliance assurance. Independent auditors typically charge fees ranging from $5,000 to $20,000 or more, depending on the complexity of the AEDT and the volume of data involved. While this is a significant expense, it is a fraction of the potential fines for non-compliance, which can accumulate rapidly if the DCWP finds violations. Moreover, an independent audit report carries more weight if the organization is ever sued for discrimination, as it demonstrates a good-faith effort to comply with the law. Employers must also consider that the auditor's independence is not just about employment status but also about financial independence; the auditor cannot have a contractual or financial relationship with the AEDT developer that could compromise their objectivity. Ultimately, for most mid-to-large employers in NYC, the investment in an independent audit is the safer and more defensible path, particularly given the DCWP's active enforcement posture in 2026.

Common Mistakes and Pitfalls in the Audit Process

Despite the clear requirements of Local Law 144, many employers stumble in their attempt to achieve compliance, often due to misunderstandings about what the audit entails or how to execute it technically. One of the most common mistakes is the misuse or misinterpretation of the 80 percent impact ratio threshold. Some employers assume that as long as their tool's overall selection rate is high, they are compliant. However, the law looks at the selection rates within protected groups, not the overall rate. If an AI tool selects 90 percent of applicants overall but does so at a rate of 70 percent for one protected group and 95 percent for another, the tool fails the audit because the impact ratio for that group is below 80 percent. Another frequent error is the use of insufficient or outdated data. The audit requires an analysis of the tool's performance, and if the data set is too small, not representative, or stale (e.g., based on hiring patterns from five years ago that no longer reflect the current labor market), the results will be unreliable. Auditors need current, robust data to draw valid conclusions. Additionally, many employers fail to ensure the auditor's true independence. Simply hiring a firm that claims to be independent is not enough; the employer must verify that the firm has no financial ties to the AI vendor and that its staff are not former employees of the company in question. A third pitfall is the failure to post the audit report publicly. The law mandates that the report be made available on the employer's website, yet some organizations either forget this step or post the report in a location that is not easily accessible to the public, such as a password-protected internal portal. This constitutes a violation of the posting requirement, regardless of whether the audit itself was technically sound. Finally, some employers attempt to use "proxy" metrics or surrogate data to hide bias, such as using zip codes as a proxy for race or using education levels that disproportionately exclude certain demographics. The DCWP is increasingly sophisticated in detecting these maneuvers, and such practices can lead to not just fines but severe reputational damage. Understanding these pitfalls is the first step toward avoiding them and achieving genuine compliance.

Practical Steps for HR and Technology Teams

For HR professionals and technology teams tasked with achieving Local Law 144 compliance, the process requires a coordinated effort that spans multiple departments. The first practical step is conducting a comprehensive inventory of all AEDTs. This should not be limited to tools that make final hiring decisions but should include any tool that influences the hiring pipeline, such as sourcing algorithms that scour LinkedIn for candidates or chatbots that pre-screen applicants via text. Once the inventory is complete, the next step is to establish a data pipeline that feeds current, accurate demographic and outcome data to the auditor. This often requires HR to clean up its applicant tracking systems (ATS), ensuring that demographic data is collected voluntarily and stored securely in compliance with privacy laws like the GDPR or CCPA, while still being usable for bias audits. HR teams should also work with their legal counsel to draft the public posting language. The posted report must be clear, concise, and accessible, explaining in plain language what the AEDT does, the results of the bias audit, and the steps being taken to address any identified disparities. Technology teams, meanwhile, must ensure that the AEDT's internal logic is transparent enough to allow for an audit. "Black box" algorithms that cannot be explained or interrogated pose a significant risk, as an auditor cannot evaluate bias if they cannot understand how the algorithm reaches its decisions. In some cases, employers may need to work with their AI vendors to obtain model explainability reports or feature importance plots that shed light on the decision-making process. Another practical step is scheduling the audit well in advance of the one-year deadline. Given the complexity of some AI systems and the potential need for data remediation, waiting until the last minute to hire an auditor is a recipe for non-compliance. Employers should aim to have the audit completed and the report posted at least three months before the AEDT's deployment date. This buffer allows time to address any issues uncovered during the audit, such as retraining the model on less biased data or, in extreme cases, discontinuing the use of the tool. Regular training for HR staff on the implications of AI bias and the specific requirements of Local Law 144 is also advisable, ensuring that everyone involved in the hiring process understands the legal and ethical stakes.

Cost, Pricing, and Resource Allocation

The financial implications of complying with NYC Local Law 144 vary significantly depending on the size of the employer, the number of AEDTs in use, and the complexity of the algorithms involved. For small businesses with a single, relatively simple hiring tool, the cost of a compliance audit might be on the lower end, potentially ranging from $3,000 to $7,000. This would typically cover the auditor's time to review the tool's documentation, analyze the data, and produce the required report. However, for mid-sized or large enterprises, the cost escalates quickly. A company using multiple AEDTs for different roles—such as one tool for executive recruitment, another for technical screening, and a third for hourly wage hiring—can expect to pay between $10,000 and $50,000 or more for a comprehensive audit program. These fees cover not just the initial bias assessment but also any follow-up analyses, data validation processes, and the preparation of public-facing reports. Additionally, there are indirect costs associated with compliance. If the audit reveals that an AEDT has a disparate impact on a protected group, the employer may need to invest in retraining the AI model, which can be a costly and time-consuming process involving data scientists and engineers. In some cases, the employer may decide that the tool is too biased to be used and must invest in purchasing a new, compliant system. Beyond the direct monetary costs, there is also the resource cost of internal management. HR staff must dedicate time to data preparation, coordination with the auditor, and managing the public posting requirement. There is also the potential cost of legal counsel to ensure that the organization's practices are fully aligned with both the letter and spirit of the law. While the upfront costs of compliance can be daunting, they must be weighed against the far greater costs of non-compliance. Fines of $1,500 per day per violation can accumulate to millions of dollars if a tool is used illegally for an extended period. Furthermore, the reputational damage of being found to use biased AI in hiring can lead to talent acquisition difficulties, employee turnover, and potential lawsuits under the New York City Human Rights Law. In the current regulatory climate of 2026, proactive compliance is not just a legal obligation but a strategic business imperative.

When to Act: Enforcement Timeline and Future Developments

Understanding the enforcement timeline of NYC Local Law 144 is critical for employers who are either newly adopting AI tools or have been using them for some time. The law became effective on July 5, 2023, and the audit requirement applies to any AEDT used after that date. However, there is a nuanced transition period that employers must navigate. If an employer was already using an AEDT before the law's effective date, they were required to conduct their first audit no later than one year after the law's effective date, meaning by July 5, 2024. For tools implemented after July 5, 2023, the audit must be completed no more than 12 months prior to the tool's use. As of September 2026, the DCWP has been actively enforcing the law, conducting investigations, and issuing fines to employers who fail to produce the required audit reports or who use tools that fail the bias threshold. Employers who have not yet complied are strongly advised to act immediately, as the likelihood of enforcement action increases with time. Looking forward, the regulatory landscape for AI in employment is evolving rapidly. NYC is not the only jurisdiction moving in this direction; states like Illinois have enacted similar laws, and the federal government is exploring AI accountability measures. There is also ongoing discussion about lowering the 80 percent threshold or introducing additional metrics to capture more subtle forms of bias. Employers should view Local Law 144 not as a one-time checkbox to be ticked in 2023 or 2024, but as an ongoing compliance obligation that will require periodic re-auditing, especially if the AI tool is updated, retrained, or if the company's hiring demographics shift significantly. Staying ahead of these developments requires a commitment to continuous monitoring and a willingness to adapt HR practices as the legal framework matures.

Conclusion: The Imperative of Algorithmic Accountability

NYC Local Law 144 has fundamentally altered the landscape of AI-driven hiring, introducing a formal mechanism for algorithmic accountability that did not exist previously. The law's audit checklist—encompassing the identification of AEDTs, the execution of an independent bias audit, the calculation of impact ratios, and the public posting of results—is a rigorous process that demands technical expertise, legal awareness, and organizational commitment. For employers, the stakes are high: non-compliance risks substantial financial penalties, legal liability, and significant reputational harm in an era where diversity, equity, and inclusion are paramount to brand value. For technology providers, the law necessitates a shift toward more transparent, explainable AI models and a willingness to subject their tools to independent scrutiny. The 80 percent impact ratio threshold serves as a clear, measurable standard, but the reality of algorithmic bias is often more subtle, requiring auditors to look beyond simple percentages to understand the systemic factors at play. As we move further into 2026 and beyond, the integration of AI into HR functions will only deepen, making compliance with laws like Local Law 144 not just a legal necessity but a competitive advantage. Organizations that embrace transparency, invest in robust auditing processes, and prioritize fairness in their AI systems will be better positioned to attract top talent, maintain employee trust, and avoid the costly pitfalls of regulatory enforcement. The definitive answer to achieving compliance is not found in a simple checklist alone, but in a holistic approach that treats algorithmic fairness as a core component of organizational governance.

FAQ

Q: What exactly is an Automated Employment Decision Tool (AEDT) under Local Law 144? A: An AEDT is any computational process, including machine learning, statistical modeling, AI, or rule-based systems, that is used to screen candidates, rank applicants, or make employment decisions such as hiring, promotion, or termination. This definition broadly covers resume screening software, video interview analysis tools, and even some scheduling algorithms, provided they assist in employment decisions.

Q: Can we use our internal HR team to conduct the bias audit, or must we hire an external firm? A: The law requires that the audit be conducted by an independent auditor. An independent auditor cannot be an employee of the employer, the developer of the AEDT, or someone with a financial interest in the tool's success. Using internal staff would likely violate the independence requirement and expose the company to enforcement action.

Q: What happens if our AEDT fails the 80 percent impact ratio threshold? A: If the impact ratio falls below 80 percent for any protected group, the law presumes disparate impact. The employer must then demonstrate that the tool is job-related and consistent with business necessity. If this cannot be proven, the employer must either discontinue the tool or implement modifications to reduce bias before using it.

Q: How often must we re-audit our AEDTs? A: The law requires an audit no more than 12 months before the AEDT is first used. While the law does not explicitly mandate periodic re-auditing after deployment, any significant update or retraining of the AI model should trigger a new audit to ensure continued compliance.

Q: Where must the audit report be posted, and what information must it contain? A: The audit report must be posted on the employer's website, specifically on the careers or jobs page. It must contain a summary of the audit results, including the impact ratios for each protected group, the methodology used, and a statement of whether the tool passes or fails the bias threshold.

Quick Facts

{ "label": "Category", "value": "NYC Local Law 144 Audit Requirement" }, { "label": "Timeline", "value": "Audit must be completed no more than 12 months before AEDT deployment; first audit deadline was July 5, 2024 for pre-existing tools." }, { "label": "Cost", "value": "Ranges from $3,000 for simple tools to $50,000+ for enterprise-wide audits of multiple AEDTs." }, { "label": "Best for", "value": "Mid-to-large employers in NYC using AI in hiring; technology providers must ensure their tools are audit-ready." }, { "label": "Penalty", "value": "Up to $1,500 per violation per day of non-compliance." } }

follow_up_keyword

nyc local law 144 compliance 2026