As employers have increasingly used artificial intelligence (AI) and other Automated Employment Decision Tools (AEDTs) to assist human resource professionals and hiring managers in reviewing voluminous resumes, federal and state governmental authorities have become concerned that these tools are disproportionately screening out female and minority applicants.

To address the adverse impact that AI and AEDTs may be causing in violation of federal and state anti-discrimination laws, state and local legislatures across the country have begun enacting and/or proposing legislation and regulations to combat these issues. New Jersey’s Legislature recently introduced a pair of bills in the Assembly to address the potential discriminatory impact the use of such technology may have on prospective job candidates.

New Jersey Assembly Bill 3854

On February 22, 2024, A. 3854 was introduced in the New Jersey Legislature with the stated purpose of regulating the use of AEDTs during the hiring process to minimize employment discrimination.

AEDTs are defined broadly in A. 3854 to mean “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 filter 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.” This definition would encompass AI programs used to filter and screen prospective job candidates.

The proposed legislation would prohibit the sale of any AEDTs in New Jersey unless:

1. The AEDT is the subject of a bias audit conducted in the year prior to the sale of the tool or the offering of it for sale.
2. The sale of the AEDT includes, at no additional cost, an annual bias audit service that provides the results of that audit to the purchaser.
3. The AEDT is sold or offered for sale with a notice stating that the AEDT is subject to the provisions of this act.

A “bias audit” is defined as “an impartial evaluation, including but not limited to testing, of an [AEDT] to assess its predicted compliance with the provisions of the ‘Law Against Discrimination,’ P.L.1945, c.169 (C.10:5-1 et seq.), and any other applicable law relating to discrimination in employment.”

In addition to requirements concerning the sale of AEDTs, A. 3854 provides that New Jersey employers who use AEDTs to screen candidates must post a summary of the most recent bias audit on their websites.

Further, the proposed legislation would require any person or entity that uses AEDTs to screen a candidate to notify the candidate of the following within 30 days:

1. That an AEDT, subject to an audit for bias pursuant to this act, was used in connection with the candidate’s candidacy.
2. The specific job qualifications or characteristics that an AEDT was used to assess in the candidate.
3. Upon request of the candidate, the source of the data collected and the data retention policy of the employer.

Violations are punishable by civil penalties of $500 for the first violation, and each additional violation occurring on that same day, and a penalty of between $500 and $1,500 for each subsequent violation. Failure to give the required notice to job candidates within 30 days would be a violation, and every subsequent 30-day period in which notice was not provided would be considered a separate violation.

Although Assembly Bill 3854 covers a number of topics relating to employers’ use of AEDTs, there are some significant unknowns including whether, and under what circumstances, the underlying bias audit will be subject to full disclosure and who may perform the required bias audit.

However, if New Jersey follows NYC’s recently enacted Local Law 144 (which already requires bias audits of AEDTs), then employers will be required to use independent auditors to perform the audit. Under NYC Local Law 144, independent auditors are persons or groups who (i) were not involved in the use, development or distribution of the AEDT, (ii) do not have an employment relationship with the employer or employment agency that seeks to use the AEDT, or the vendor who developed or distributed the AEDT, and (iii) do not have a direct financial interest, or material indirect interest, in the employer, employment agency or vendor.

Regardless of the parameters of permissible auditors, it would be prudent for employers, and vendors alike, to consult with attorneys who are experienced with adverse impact studies to assist in developing, conducting and summarizing the results of the bias audit. This is particularly important to enable employers and vendors to not only obtain their insight regarding relevant state and federal anti-discrimination laws, but also to potentially obtain confidentiality of the bias audit process through the attorney-client privilege.

New Jersey Assembly Bill 3911

A. 3911 was introduced on February 27, 2024. It seeks to regulate the use of AI analysis of job applicants’ recorded video interviews that are submitted to employers. Pursuant to the proposed legislation, New Jersey employers that request applicants to record video interviews and then use an AI program to analyze the applicant-submitted video must:

1. Notify an applicant before the interview that AI may be used to analyze the applicant’s video interview and consider the applicant’s fitness for the position.
2. Provide an applicant with information before the interview explaining how the AI works and what general types of characteristics it uses to evaluate applicants.
3. Obtain, before the interview, written consent, which may be electronic, from the applicant to be evaluated by the AI program as described in the information provided.

An employer must take these steps prior to requesting the applicant submit a video interview.

Significantly, the proposed legislation provides that New Jersey employers who use AI analysis of recorded video interviews must collect data concerning (i) the race and ethnicity of applicants who are and are not afforded the opportunity for an in-person interview after the use of AI analysis; and (ii) the race and ethnicity of applicants who are offered a position or hired.

However, the proposed legislation does not provide the manner or method in which this demographic information is to be collected from applicants. Ideally, this required information would be collected from the applicant prior to the recorded video interview via voluntary self-identification, otherwise employers will have to use less accurate methods such as visual identification.

The collected demographic data must then be reported annually by the employer to the New Jersey Department of Labor and Workforce Development (NJ DOL). The NJ DOL would be responsible for analyzing the demographic data and reporting to the Governor and the Legislature each year whether the data discloses a racial bias in the use of AI.

If the law passes, employers should perform a disparate impact analysis of the data gathered prior to submission to the DOL to enable them to determine whether adverse impact exists, and, if so, what steps can be taken to remediate any biases during the hiring process. As with the bias audit requirement proposed by A. 3854, employers should consult with experienced attorneys before conducting the disparate impact analysis. They can assist in conducting the review and potentially shield it through the attorney-client privilege.

Under the proposed legislation, employers who request applicants to submit video recorded interviews would not be permitted to share an applicant’s video except with a service provider whose expertise or technology is necessary to evaluate the applicant’s fitness for a position. Further, upon receipt of a request from an applicant to delete the video recorded interview, employers would have 30 days to comply and instruct any other persons who received copies of the applicant’s video interview to do the same. This could be problematic for reporting purposes if the demographic data is not collected in advance. Moreover, it is unclear if the video recorded interview data could be used to evaluate the candidate if they provide their consent and later request the video be deleted.

Violations of the act would carry civil penalties of $500 for a first offense and $1,000 for any subsequent offense.

Takeaways

The foregoing legislation has merely been proposed by the New Jersey State Legislature at this time. Still, employers who use, or are contemplating using, AEDTs in their hiring process should be mindful of the pending legislation and contemplate procedures and policies for compliance should it be enacted. Regardless of whether either bill is enacted, it is likely that New Jersey will at some point in the not too distant future follow the lead of other jurisdictions, such as Illinois and New York City, and enact laws aimed at regulating employers’ use of AEDTs in the hiring and/or employment decision making process. Other states including, Vermont, California, and Massachusetts, have similarly proposed laws to address the use of AEDTs.

Employers should be prepared for this reality and take steps to affirmatively ensure any AEDTs currently in use are not having a discriminatory impact on any protected class in violation of state or federal laws.

[View source.]