SENATE LABOR COMMITTEE

STATEMENT TO

SENATE, No. 4075

STATE OF NEW JERSEY

DATED: OCTOBER 1, 2026

      The Senate Labor Committee reports favorably Senate Bill No. 4075.

      This bill is designed to ensure that there are reasonable guard rails on the use of systems using artificial intelligence (AI) and related technologies in places of employment and for public services.  The bill regulates AI used in electronic monitoring tools (EMT) to collect and process a wide range of data and information about workers and job applicants and AI used by automated employment decision-making systems (AEDS) in making employment-related decisions such as hiring and firing.  The bill also regulates automated benefit and service decision-making systems (ABSDS) used in making decisions related to providing public benefits and services.

The need to regulate AI-based systems

      Unregulated automated systems in places of employment are not held accountable.  Information from hundreds of sources, often inaccurate or incorporating longstanding biases and prejudices, can have serious adverse economic impacts on workers and claimants for benefits and services.  Affected individuals rarely have access to the information used or any explanation of decision processes that can have such profound effects on their lives.

      AI technologies have the potential to be useful tools to improve human decision-making, but only if both the people making the decisions and the people affected by the decisions have active, well-informed roles in determining how the tools are used.  Because the use of AI is also another form of automation with obvious potential to displace workers, the implementation of AI technologies should be subject to requirements that employers address concerns raised by workers and unions, to seek a balance between the needs of employers for productive efficiency and the needs of workers to stay productively engaged and be involved in how the technology is implemented.  Moreover, the established rights of workers to organize and bargain for a just share of the benefits of increased productivity must be protected from the threat of invasive AI-driven surveillance and intimidation.

Prohibited uses of AI-based systems

      This bill requires that an AI system, specifically an EMT, AEDS, or ABSDS, cannot be used until it is screened in an “impact assessment” and approved as being in compliance with the requirements of the bill by an independent auditor or, if for governmental use, by the Department of Labor and Workforce Development.  Compliance with the bill includes prohibitions of:

      (1)  Surveillance practices such as monitoring workers in bathrooms, rest areas, locker rooms or other non-work areas or outside of the workplace, surveillance outside of work times, or compelling workers to install monitoring software in personal devices, have tracking devices implanted, or disclose device passwords or social media accounts;

      (2)  Using any AI system or surveillance in a way that causes, or contributes to, a violation of a collective bargaining agreement or of a State or federal law, or pressures employees not to exercise their rights under an agreement or those laws;

      (3)  Using an AI system to set, or facilitate setting, performance standards or productivity quotas likely to contribute significantly to harming worker health and safety; and

      (4)  Using AI systems to discriminate in employment decisions, including hiring or firing, or decisions about public benefits or services, by using information regarding any classification or characteristic of an individual protected under State or federal law, or information about union affiliation, or any other classification or characteristic, other than unlawful behavior, not directly related to work performance or work qualifications or eligibility for benefits or services.

Procedures for impact assessments and approvals of AI-based systems

      The impact assessment for an AEDS or EMT must also confirm that it is primarily intended to be used for specified business purposes like ensuring the quality of goods and services, assessing employee performance, administering wages and benefits, and complying with relevant laws, all done with means least invasive to workers, without compiling or using data not needed for the specified purposes.

      Vendors of AEDS’s seeking approval for their use must provide the independent auditor or the department conducting the impact assessment with:

      (1)  Detailed documentation about AEDS design and development, and sources of the data used to develop and train the AEDS;

      (2)  Its intended purpose, deployment context, rationale for use, and its outputs and types of employment decisions in which the outputs will be used;

      (3)  An analysis of the accuracy, reliability, validity, and error rates of the AEDS, including the reasonably foreseeable effects of tuning, retraining, or modification; and

      (4)  What impacts its use may have on efficiency and output for the employer that deploys it, including quantified estimates of: the increased earnings and cost savings for the employer, any employment losses for the workers, any new employment related to human oversight, and the percent of the increased earnings and cost savings that are due to reduced employment.

      The full report of the impact assessment, and a summary of the report, produced by the independent auditor or the department, is to be kept in a public registry maintained by the department and made available to employees and their unions, and employers or public entities using the AEDS, ABSDS, or EMT.

Notification and disclosure requirements; right to have employers respond to worker and union concerns

      The employer or public entity is required to give written notice to its employees and any union representing them, 60 days in advance, that it intends to implement the AEDS or EMT.  The notice is required to include:

      (1)  A detailed description of the AEDS, ABSDS or EMT and how they will be used;

      (2)  A copy of the summary of the impact assessment report and instructions for accessing the entire report from the public registry;

      (3)  Disclosure of the right to have access to all information relevant to contest any disclosure in the notice;

      (4)  A statement of the obligation of the employer or entity to address concerns raised by any union of their employees over any possible lack of compliance with the requirements of the bill; and

      (5)  Descriptions of any performance standards or productivity quotas used to evaluate workers.

      Employers are not required to have performance standards or productivity quotas, but they are not allowed to take any adverse action against an employee for not meeting a standard or quota that was not disclosed.

      An employer or public entity is required to notify employees of any significant change in the AEDS, ABSDS or EMT or the way they are used, and provide the same opportunity to raise concerns regarding potential violations as in the initial implementation.  The bill requires employers and vendors to keep all information about individuals accurate and up to date, and give them access to the information and a right to seek corrections of inaccurate information or information kept in violation of the bill’s requirements, either when notice is given of a change, or at least once per year even if there is no change.

Data security

      The bill prevents the release of information by an employer, public entity or vendor to third parties without the individual’s consent in most circumstances, and requires removal of information when the legitimate need for it ends.  Employers and vendors are responsible for data security and are jointly and severally liable for damages caused by any security breach.

Procedures and appeal rights regarding adverse decisions with AI-based systems

      Employers and public entities are required to notify individuals of adverse decisions like a termination or a denial of public benefits or services at least 10 days in advance and give affected individuals and their authorized representatives an explanation of the decision, access to all relevant information, and the right to contest the decision through a discussion with a human reviewer employed by the employer or entity and given authority to correct errors, or, if the individual not satisfied, through an administrative or civil action or a grievance or arbitration procedure under a collective bargaining agreement.

Human oversight of AI-based systems

       Employers and public entities are not allowed to rely primarily on data or outputs from an EMT, AEDS, or ABSDS, or information from third parties like data brokers, when making decisions about employment, benefits or services, or to make decisions based exclusively on EMT, AED or ABSDS information or data.  The bill requires users of those systems to establish meaningful human oversight of them, designating employees as internal reviewers with sufficient expertise about the systems to identify potential errors, biases, or inaccuracies produced by their use, and authority to revise or reject system outputs.

      The bill also gives all employees the right to not follow AI outputs which they reasonably believe will cause harm, after notifying the employer of the problem if there is time to do so.

Public employer use of AI-based systems and technologies reducing employment a mandatory subject of bargaining; rights of displaced workers

      The bill makes a public employer decision to implement an EMT, AEDS or ABSDS or a new technology likely to result in a significant job loss a mandatory subject of collective negotiation.  The public employer may not implement any of those systems or technologies during the term of a current contract, but must, at least 30 days before the contract expires, give at least 90 days’ notice before the date of the implementation and negotiate in good faith over the impact.  Public employees who are displaced as a result are entitled to:

      (1)  Receive severance pay and health benefits for least one year, or at least two years for employees with five or more years of service;

      (2)  Retain acquired seniority rights;

      (3)  Recall rights; and

      (4)  The right to any training needed to fulfill human oversight roles and human roles dealing with problems individuals have with AI decisions, and priority for any new employment in those roles.