How to Create a Compliant NYC Local Law 144 Bias Audit Report
Step-by-Step Guide
New York City’s Local Law 144 (LL 144) regulates the use of automated employment decision tools (AEDTs) in hiring and promotion decisions. Since enforcement began in July 2023, employers and employment agencies operating in NYC are required to undergo an annual independent bias audit and publish a report that complies with detailed transparency requirements.
This guide walks through exactly what a Local Law 144 bias audit report must include, how the required metrics are calculated, and how to format and publish the results to meet the City’s standards.
What Is the Purpose of a LL 144 Bias Audit Report?
The bias audit report serves two primary legal functions:
To demonstrate that an independent bias audit has been conducted within the past 12 months for any AEDT used to screen candidates or employees for hiring or promotion.
To make key audit results publicly accessible on the employer's website, ensuring transparency for job candidates and regulatory bodies.
The report must be published before the AEDT is used and must remain accessible for at least six months after the tool’s last use.
Step 1: Confirm Whether LL 144 Applies to You
Before creating a report, ensure that your situation falls under the scope of LL 144. The law applies if all of the following are true:
You use an Automated Employment Decision Tool (AEDT) that employs machine learning, statistical modeling, data analytics, or AI to assist or replace human discretion in employment decisions.
The AEDT is used to screen candidates or employees for hiring or promotion.
The job is located in New York City, associated with an NYC-based office, or the employment agency is based in NYC.
Step 2: Conduct the Required Bias Audit
A bias audit must be performed by an independent auditor, which is defined as a party with no involvement in developing, distributing, or using the AEDT and no financial interest in the employer or vendor.
There are two types of audits depending on how the AEDT operates:
A. Selection-Based Audit
If the AEDT is used to make yes/no decisions (e.g., selecting candidates for interviews or classifying them as qualified/unqualified), the audit must include:
No of applicants in each group
No of selected applicants in each group
Selection rate for each demographic group
Impact ratio for each group compared to the group with the highest selection rate
B. Scoring-Based Audit
If the AEDT assigns scores to candidates (e.g., 0–100 fit rating), the audit must include:
No of applicants in each group
Scoring rate: % of individuals in each group who scored above the sample median
Impact ratio: Scoring rate for each group divided by the highest scoring rate
For both types, the audit must disaggregate results by:
Sex categories
Race/Ethnicity categories
Intersectional categories (combinations of sex and race/ethnicity)
Groups with fewer than 2% of the total audit population may be excluded from impact ratio calculations but must still be listed with raw counts and scoring/selection rates.
Step 3: Structure the Report: Required Elements
Below is a comprehensive list of the elements that must appear in the publicly available report, as required by NYC law.
1. Cover Information
Include the following identifiers:
Employer or employment agency name
AEDT model name and version
Distribution date (when the AEDT was first used)
Bias audit date (when the most recent audit was completed)
Independent auditor’s name or firm
2. Description of the AEDT
Provide a brief summary of:
The purpose of the tool (e.g., resume screening, interview scoring)
Whether the audit used historical data (real-world applicants previously screened) or test data
A plain-language explanation of the data used
If test data is used, explain why historical data was insufficient and how the test data was generated.
3. Data Summary
Disclose:
Time range covered by the audit data (for historical data)
Total number of individuals assessed
Number of individuals in “unknown” demographic categories
These are candidates for whom race/ethnicity or sex was not provided
These individuals are excluded from impact-ratio calculations but must still be counted
4. Results Tables
Present results in three separate tables:
A. Sex Categories
| Sex | # Candidates | # Selected / Scoring Rate | Impact Ratio |
B. Race/Ethnicity Categories
| Race/Ethnicity | # Candidates | # Selected / Scoring Rate | Impact Ratio |
C. Intersectional Categories
| Sex | Race/Ethnicity | # Candidates | # Selected / Scoring Rate | Impact Ratio |
For each group, indicate:
The reference group (with the highest rate, used as denominator for impact ratio)
A dash (-) for the impact ratio if the group was excluded under the 2% rule
5. Exclusion Notes (If Applicable)
If any categories were excluded due to the 2% threshold, the report must:
Identify the group(s) excluded
State the number of applicants in that group
Provide the selection or scoring rate
Explain why the exclusion occurred
6. Auditor Independence Statement
Include a declaration that the auditor:
Is independent and impartial
Was not involved in developing, using, or profiting from the AEDT
Has no financial interest in the employer or the vendor
Step 4: Publish the Report
The final report must be made publicly available in a clear and conspicuous manner:
Post it on the careers section of your company’s or agency’s website
You may use a direct hyperlink labeled clearly as “Bias Audit Results” or similar
Keep the report accessible for at least 6 months after the AEDT’s last use
Step 5: Don’t Forget Candidate Notifications
In addition to the public report, LL 144 requires that candidates and employees receive advance notice if an AEDT will be used in their assessment.
This notice must:
Be delivered at least 10 business days in advance
Include a description of the characteristics the AEDT evaluates
Explain how individuals may request a reasonable accommodation or alternative process (if available)
You may deliver this notice via:
Job postings
Email or mail
Website notices (for job applicants)
Internal policies (for current employees)
Make LL 144 Painless with BiasBeacon
Manually building a compliant bias audit report can be time-consuming, error-prone, and legally risky. That’s why we built BiasBeacon — a simple SaaS platform that automates the entire process.
Just upload your model-score CSV, and within seconds, BiasBeacon generates a DCWP-compliant PDF report that includes:
Selection or scoring rates by demographic group
Impact ratios (including sex, race/ethnicity, and intersectional categories)
Count of unknown demographic entries
Audit date and AEDT distribution date
Required exclusions and explanations
Auditor Independence Statement
The first audit is free, so you can test your data and publish a legally sound report in minutes — no spreadsheet gymnastics, no legal guesswork.
➡ Try it today at BiasBeacon and generate your audit in seconds.
Final Thoughts
Creating a compliant LL 144 bias audit report is not just about checking legal boxes—it’s a public statement of your commitment to transparency and fairness in the hiring process. While the law does not require you to act on the audit results, ignoring clear disparities can lead to scrutiny under broader anti-discrimination laws.
A well-prepared report is:
Clear
Complete
Legally defensible
Easy for candidates and regulators to understand
Invest in the process, document everything, and treat the audit as an opportunity to improve—not just comply.
