New Jersey AI Legislation in 2026

The New Jersey Picture

New Jersey has chosen enforcement over comprehensive statute. As of June 2026 the state has no omnibus private-sector AI law on the books, and I want to be precise about that, because the absence is itself the story. Rather than wait for a sweeping framework, New Jersey activated tools it already had. The most consequential move was administrative, not legislative. In January 2025 Attorney General Matthew Platkin and the Division on Civil Rights declared that the decades-old Law Against Discrimination already reaches algorithmic discrimination. That single interpretive act means any employer, landlord, lender, or business using an automated decision tool in New Jersey is already exposed to liability under existing civil rights law today, with no new statute required. Crucially, that liability can attach without discriminatory intent and even when a third-party vendor built and trained the model.

The second pillar is criminal. With A3540, enacted as P.L.2025 chapter 40 on April 2, 2025, New Jersey put real teeth behind the misuse of generative media. Producing or disseminating a malicious deepfake is now a third-degree crime carrying up to five years of imprisonment and fines up to $30,000, and victims gained a private right to sue. This is a distinctly New Jersey posture. The deepfake law folds AI-generated media into existing offenses rather than building a standalone AI code, and it was driven by a real harm to a New Jersey teenager rather than by abstract policy. The law preserves space for satire, journalism, teaching, and research, but the signal to builders of synthetic media tools is unambiguous.

Everything else remains in motion. The automated-employment-decision-tool bills from the 2024-2025 session, including A3854 with its mandatory annual bias-audit regime, did not reach the Governor's desk, and the 2026 successor effort, S1840, the Responsible AI Advancement and Workforce Protection Act, is only just introduced and sitting in the Senate Labor Committee. The companion-chatbot and mental-health disclosure bills are likewise pending. None of these is law yet, and I would caution against treating any of them as compliance obligations today. They do, however, telegraph where New Jersey is heading: bias audits, high-risk system impact assessments, human-disclosure requirements for chatbots, and worker-protection mandates tied to AI-driven layoffs.

What should builders and businesses do now. First, treat the Law Against Discrimination guidance as live law, not aspiration. Inventory every automated tool touching hiring, promotion, housing, credit, or customer access, demand bias documentation from vendors, and run your own outcome testing, because vendor reliance is not a defense in New Jersey. Second, if you generate, host, or distribute synthetic audio or video, build provenance, disclosure, and consent controls now to stay clear of the criminal statute. Third, watch S1840 and the chatbot bills and pre-build the obvious requirements, namely impact assessments for high-risk systems and clear you are talking to AI disclosures, so that enactment becomes a switch you flip rather than a project you start.