NYC Publishes Final Rule for AEDT Law and Identifies New Enforcement Date

On April 6, 2023, the New York City Department of Consumer and Worker Protection (“DCWP”) issued a Notice of Adoption of Final Rule to implement Local Law 144 of 2021, legislation regarding automated employment decision tools (“AEDT Law”).  DCWP also announced that it will begin enforcement of the AEDT Law and Final Rule on July 5, 2023.   Pursuant to the AEDT Law, an employer or employment agency that uses an automated employment decision tool (“AEDT”) in NYC to screen a candidate or employee for an employment decision must subject the tool to a bias audit within one year of the tool’s use, make information about the bias audit publicly available, and provide notice of the use of the tool to employees or job candidates.

The adoption of the Final Rule followed the DCWP’s previous proposals in September and December 2022 and a review of public comments.  The Final Rule adopts the December 22 proposal, with modest amendments, and by way of summary:

1. defines terms used in the AEDT Law;
2. clarifies how bias audits can comply with the AEDT Law;
3. clarifies the requirements for the published results of the required bias audit;
4. reiterates the requirements for notices that employers and employment agencies must provide to employees and job candidates; and
5. clarifies other obligations employer or employment agencies must comply with.

Definitions

Pursuant to the Final Rule, the phrase “to substantially assist or replace discretionary decision making” means: (1) “to rely solely on a simplified output (score, tag, classification, ranking, etc.), with no other factors considered”; (2) “to use a simplified output as one of a set of criteria where the simplified output is weighted more than any other criterion in the set”; or (3) “to use a simplified output to overrule conclusions derived from other factors including human decision-making.”  

The Final Rule also provides a broad interpretation of the phrase “machine learning, statistical modelling, data analytics, or artificial intelligence.”  Notably, the Final Rule does not include a previously proposed limitation that “inputs and parameters are refined through cross-validation or by using training and testing data.” 

For purposes of the bias audit requirement, the Final Rule clarifies that an “independent auditor” is a person or group that is capable of exercising “objective and impartial judgment on all issues within the scope of a bias audit.”  The “independent auditor,” cannot be involved in using, developing, or distributing the tool, have an employment relationship with an employer or employment agency that seeks to use the tool during the bias audit, or have a direct financial interest or a material indirect financial interest from the tool’s use during the bias audit.

Bias Audit Compliance

The Final Rule states that a bias audit must calculate the impact ratio of each EEO protected category, including sex, race/ethnicity, as well as intersectional categories of sex, ethnicity, and race.  The impact ratio, which may exclude an EEO protected category that comprises less than 2% of the data being used, must be calculated in one of two ways: (1) as a selection rate for a category divided by the selection rate of the most selected category; or (2) the scoring rate for a category divided by the scoring rate of the highest scoring category. 

In addition, as a general matter, the Final Rule explains that the bias audit must use data from the employer’s or employment agency’s own historical use of the AEDT.  Where, however, this historical data is insufficient – either because the employer or employment agency has never used the AEDT, or because its historical data is not statistically significant to conduct a bias audit – the independent auditor may use historical data of other employers or employment agencies.  As an alternative, an employer or employment agency can use “test data.” In that circumstance, the bias audit must summarize why historical data could not be used.

Publishing Audit Results

The Final Rule clarifies that published results should include the distribution date of the AEDT and must contain an active hyperlink to a website that contains the bias audit summary and the tool’s distribution date.  

Notices to Employees and Job Candidates

The Final Rule reiterates the requirement that employers and employment agencies notify candidates at least 10 business days before using AEDT.

In preparation for the July 5, 2023 enforcement date, each employer and employment agency using AEDTs should consider taking the following steps:

• Analyze and determine whether its use of an AEDT triggers compliance with the AEDT Law;
• If using AEDT in manner requiring AEDT Law compliance;
  • Identify and gather data to conduct a bias audit;
  • Retain an independent auditor to conduct a bias audit;
  • Consider where and how to publish the results of the bias audit; and
  • Ensure compliance with all notice requirements.

Epstein Becker & Green, P.C. will continue to monitor and report on developments regarding the AEDT Law and all similar workplace laws and regulations.  In the meantime, please feel free to contact the authors if you have any questions regarding complying with the AEDT Law or would like to inquire about the use and regulation of workplace artificial intelligence tools in general.