The New Jersey’s Senate Labor Committee voted 5-0 on Oct. 1, 2026, to refer a bill to regulate employment and other uses of AI-based technologies to the Senate Budget and Appropriations Committee.

The “Fair AI Regulations (FAIR) Act” (S4075) seeks to bar discriminatory use of AI by regulating three categories of technology:

  1. Electronic monitoring tools (EMTs), which collect data about workers and applicants. 
  2. Automated employment decision systems (AEDS), which are used in decisions such as hiring and firing. 
  3. Automated benefit and service decision systems, which public entities use to decide on public benefits.

The bill defines AEDS broadly to cover any computational process that uses machine learning, statistical modeling, data analytics, AI, or similar methods to produce outputs an employer uses in employment-related decisions that affects an employee’s “terms or conditions of employment.” “Terms and conditions of employment” covers wages, schedules, performance evaluations, discipline, promotion, and work assignments.

If enacted in its current form, S4075 would take effect 18 months after it becomes law and significantly increase employers’ compliance burden.

Key Employment Requirements

  • Pre-deployment impact assessments. An employer may not use an AEDS or EMT until an independent auditor (or, for public employees, the state Department of Labor) has completed an impact assessment. The assessment must confirm the tool serves an allowable purpose, such as quality control, performance assessment, legal compliance, safety, or wage administration. It must also confirm the tool uses the least invasive means and the least data necessary. For tools already in use, the assessment is due within six months of the Act’s effective date. Assessment reports go into a public registry maintained by the Department.
  • Notice. Employers must give employees and any union written notice at least 60 days before implementation of the tool, and another 60 days’ notice before any significant change. The notice must describe any productivity quotas or performance standards. An employer may not take adverse action based on an undisclosed quota. If a union raises concerns within 30 days of notice, the employer may not implement the tool until it has responded in writing.
  • Human oversight and prohibited practices. No employment decision may rest solely or determinatively on AI outputs or monitoring data. Employers must designate trained internal reviewers with authority to reject those outputs. The bill also bars:
    • Monitoring employees while they are off duty or on breaks;
    • Requiring monitoring software on employees’ personal devices; and
    • Taking adverse action based solely on keystroke or idle-time tracking.
  • Data obligations. Employers must keep records for three years and destroy the data within 37 months unless the employee consents to longer retention. They must notify affected employees and the Department of a data breach within 48 hours.

Employee Rights

Employees may see and correct the data held about them at least once a year. Employers must give employees at least 10 days’ written notice explaining the decision and how the system was used before an adverse AI-assisted decision takes effect, and, within 30 days, employees may review the underlying data and to appeal to a designated human reviewer with authority to overturn the decision.

The Act also protects certain employees from retaliation for refusing to follow AI outputs they reasonably believe could cause harm. The employee must have raised the concern with a supervisor, and the employer failed to adjust the output or give a reasonable written explanation or the risk of harm was too urgent to await correction. Any waiver of rights under the Act is void.

Enforcement; Penalties

A knowing and willful violation is a disorderly persons offense. Fines and jail time are possible. The commissioner of labor may also impose administrative penalties. The bill creates a private right of action. Vendors and employers are jointly and severally liable for violations under the Act.

Key Concerns for Employers

The New Jersey Law Against Discrimination already prohibits both intentional discrimination and employment practices that have an unlawful disparate impact on protected classes, and the state’s Division on Civil Rights has indicated that these protections apply to AI tools. Accordingly, there are notable concerns for employers:

  • Scope. The Act’s definitions could reach routinely used program simply for using “data analytics” to inform a decision.
  • Information employers do not have. Businesses that buy off-the-shelf products may be unable to get the vendor’s proprietary development and training information, making compliance cumbersome.

Employers should start inventorying their workplace technology and reviewing vendor contracts in anticipation of S4075 becoming law.

Contact a Jackson Lewis attorney with any questions.