AI Frontier Institute

Regulates use of artificial intelligence-based systems for electronic monitoring regarding employment and public services.

View primary source →

New Jersey Assembly Bill 4981 (introduced May 7, 2026; referred to the Assembly Science, Innovation and Technology Committee) has not been enacted. If enacted, it would regulate AI-based and electronic systems used for workplace monitoring and for public benefit decisions. It covers three kinds of system: electronic monitoring tools (EMTs), automated employment decision systems (AEDS), and automated benefit or service decision systems (ABSDS). Section 2 would bar uses that violate collective bargaining agreements or labor law, or that harm health, safety, privacy or dignity. It would also bar monitoring in bathrooms and other private areas, and monitoring of off-duty employees. Employers could not require monitoring software on personal devices, implanted devices, or access to personal accounts. The bill would bar using protected-class or union information in decisions, and sharing or using biometric and health data in decisions. It would also bar retaliation, and adverse action based solely on keystroke, idle-time or mouse-movement tracking. Section 3 would require an independent auditor, or the Department of Labor and Workforce Development for public employees, to complete an impact assessment before an AEDS or EMT is deployed. For an ABSDS, section 4 would require the Department to do the assessment. Assessments would be filed in a public registry, and vendors would pay the Department's direct costs for the public-sector assessments. Section 6 would require 60 days' written notice to affected workers, beneficiaries and unions, and would require employers to respond to union concerns before implementing. Section 8 would require 10 days' notice of adverse decisions, access to the data, and an appeal to a human reviewer. Section 9 would bar decisions based solely on automated outputs and would require meaningful human oversight. Section 10 would protect employees who refuse harmful AI outputs. Sections 11–13 would make public-employer automated services agreements a mandatory subject of negotiation, restrict when they can be signed, and give displaced public employees severance, health coverage, seniority, recall rights and training. Section 5 would set data-use limits, retention rules, and breach notice within 48 hours. Enforcement would include criminal penalties (section 18), administrative penalties (section 18), unfair practice charges (section 14), civil actions with damages and fines (section 19), and joint and several liability for vendors and deployers (section 20). Section 22 would make the act take effect 18 months after enactment.

← Back to the tracker