As previously reported, employers’ use of artificial intelligence to assist in employment decision-making is being subjected to ever-increasing regulatory scrutiny at all levels of government―local, state, and federal. In the New Jersey General Assembly, A4909 (“A4909”) was introduced on December 5, 2022, proposing a host of legal requirements that would apply to automated employment decision tools. On January 9, 2023, New York State introduced State Assembly Bill A00567 (“A00567”), modeled after New York City’s recently passed Automated Employment Decision Tool Law, set to go into effect on April 15, 2023. If enacted, the New York State Law would prohibit the use of certain automated employment decision tools unless employers take affirmative steps to screen the technology, including running a bias audit. (Preliminary indications from the rulemaking process suggest that Rules issued under the New York City law may substantially limit its impact by narrowly defining which tools are covered.)
New Jersey: A4909
A4909 seeks to regulate the sale and use of “automated employment decision tools.” The bill defines these as “any system the function of which is governed by statistical theory, or systems the parameters of which are defined by systems, including inferential methodologies, linear regression, neural networks, decision trees, random forests, and other learning algorithms, which automatically filters candidates or prospective candidates for hire or for any term, condition or privilege of employment in a way that establishes a preferred candidate or candidates.”
A4909 would make unlawful the sale of automatic employment decision tools unless:
- the tool is the subject of a bias audit conducted in the past year prior to selling the tool or offering the tool for sale;
- the sale of the tool includes, at no additional cost, an annual bias audit service that provides the results of the audit to the purchaser; and
- the tool is sold or offered for sale with a notice stating that the tool is subject to the provisions of [A4909].
The bill defines “bias audit” as “an impartial evaluation, including but not limited to testing, of an automated employment decision tool to assess its predicted compliance with the provisions of the [Law Against Discrimination], and any other applicable law relating to discrimination in employment.”
A4909 also contains a notification requirement. Within 30 days of “any person” using the tool to screen a candidate for an employment decision, they would need to notify “each candidate” that the tool was “used in connection with the candidate’s application for employment” and that it “assessed the job qualifications or characteristics of the candidate.”
Penalties for non-compliance would include a civil penalty of up to $500 for the first violation and “each additional violation occurring on the same day as the first violation.” For every subsequent violation, penalties would be assessed between $500 and $1500. Violations would accrue daily for each automated employment decision tool sold or offered for sale in violation of A4909. The bill further provides that each instance of notice not being provided to candidate(s) within 30 days of using the tool is a single violation and that “each 30-day period thereafter in which the notice is not provided” constitutes a separate violation.
New York: A00567
A00567 defines “automated employment decision tool” more broadly than the New Jersey bill as “any system used to filter employment candidates or prospective candidates for hire in a way that establishes a preferred candidate or candidates without relying on candidate-specific assessments by individual decision-makers. Automated employment decision tools shall include personality tests, cognitive ability tests, resume scoring systems, and any system whose function is governed by statistical theory or whose parameters are defined by such systems, including inferential methodologies, linear regression, neural networks, decision trees, random forests and other artificial intelligence or machine learning algorithms.” Further, A00567 qualifies that tools that do not “automate, support, substantially assist or replace discretionary decision-making processes” and that do not “materially impact natural persons” would not be considered automated employment decision tools.
Under A00567, automated decision tool vendors would be required to conduct a “disparate impact analysis” at least annually to assess the “the actual impact of any automated employment decision tool used by any employer to select candidates for jobs within the state.” The vendor must provide the analysis to any employer seeking to use the tool, but the law emphasizes that the analysis does not need to be publicly filed and would be “subject to all applicable privileges.” Second, and before the implementation or use of the tool, employers must make publicly available a summary of the most recent disparate impact analysis, as well as “the distribution date of the tool to which the analysis applies” on the employer or employment agency’s website. Finally, on at least an annual basis, the employer must provide to the New York Department of Labor the most recent disparate impact analysis summary provided to the employer on that tool.
A “disparate impact analysis” is defined as “an impartial analysis, including but not limited to testing of the extent to which use of an automated employment decision tool is likely to result in an adverse impact to the detriment of any group on the basis of sex, race, ethnicity, or other protected class under article fifteen of the executive law.” The definition also provides that the analysis “shall differentiate” between candidates selected and not selected for a position and “shall include” an analysis following the EEOC’s Uniform Guidelines on Employee Selection Procedures.
This bill would permit the State’s Attorney General and the Commissioner of Labor to initiate investigations if a preponderance of the evidence (including the summary of the most recent disparate impact analysis) establishes their suspicion of a violation.