Proposed New Jersey Legislation Signals Expanding AI Compliance Obligations

Like many jurisdictions across the United States, the New Jersey Legislature is advancing a cluster of proposed, sector-specific bills to regulate the use of generative artificial intelligence. These bills target areas ranging from real estate advertising and election-related communications to consumer interactions, licensed professions, and AI companion technologies. Taken together, they reflect a clear legislative direction: New Jersey is approaching AI not as an abstract technological development, but as a practical compliance and governance issue across private industries and the public sector. Each of the five bills discussed below was introduced on March 10, 2026, and several are out of committee—signaling that regulatory change may be imminent.

Collectively, the bills employ distinct definitions of AI technologies and regulated actors and impose a range of compliance obligations—including disclosure requirements, use restrictions, and documentation duties—backed by varying enforcement mechanisms and penalty structures. Certain bills also contemplate further regulatory development through agency rulemaking, indicating that the current proposals may represent only the initial phase of a broader, evolving framework governing AI use in New Jersey. The following provides a concise overview of these pending bills.

A4728 — AI-Altered Real Estate Advertising

A4728 would supplement Title 45 and regulate the use of generative AI and photo-editing tools in the advertising of dwelling units and dwelling sites. As amended, it applies to properties advertised for sale, rent, or lease, and expands the definitions of “buyer” and “seller” to include prospective tenants and landlords in rental and lease transactions. A4728 § 1(a). The bill defines “generative artificial intelligence” as a technology system trained on data, designed to simulate “human connection” through text, audio, or visual communication, and capable of generating non-scripted outputs with limited or no human oversight. Ibid. This definition of “generative artificial intelligence” differs from those employed in all other pending legislation.

The bill would prohibit a seller or seller’s agent from using photographs that have been created or edited, in whole or in part, using generative AI or other photo-editing software in a manner that fundamentally alters the image. Id. at § 1(b). Advertisements must reflect the actual, current condition of the dwelling unit or dwelling site, and any image used may not be more than five years old. Ibid. The bill permits a limited exception for the addition or removal of furniture and other non-fixed items—i.e., virtual staging or decluttering—provided that the image otherwise reflects the property’s actual condition and is not used to conceal defects. Id. at § 1(c). Where such tools are used, the seller or seller’s agent must disclose that generative AI or other photo-editing software was utilized and, upon request, provide the original, unaltered images. Id. at § 1(d).

The bill outlines concrete penalties and prompt enforcement mechanisms. A violation would subject a seller to a civil penalty of $500 for a first offense and up to $1,000 for each subsequent offense, recoverable through a summary proceeding under the Penalty Enforcement Law of 1999, with jurisdiction in both the Superior Court and municipal court. Id. at § 1(e). A seller’s agent who violates the bill, or any rule or regulation adopted pursuant to it, would be subject to disciplinary action by the New Jersey Real Estate Commission. Ibid. The bill also directs the New Jersey Real Estate Commission to promulgate implementing regulations within one year of enactment. Id. at § 2. The act would take effect on the first day of the second month following enactment, permits anticipatory regulatory action, and applies prospectively to advertisements for properties first listed after the effective date. Id. at § 2. In short, the bill draws a clear line between permissible AI-assisted marketing (e.g., virtual staging or decluttering) and impermissible image manipulation, while requiring disclosure of any such use.

A4729 — Election-Related AI Chatbots

A4729 would supplement Title 19 and target AI chatbots that provide election-related information. The bill adopts a slightly different definition of “generative artificial intelligence” than that employed in A4728, describing generative AI as a data-trained system designed to simulate “human communication” through text, audio, or visual formats and to generate non-scripted outputs with limited or no human oversight. The bill also provides other, more expansive definitions tailored to its subject matter. A4729 § 1(a). “[A]rtificial intelligence chatbot” is a software application, web interface, or computer program that simulates “human conversation and interaction” through textual or aural communications. Ibid. “Candidate” expressly includes individuals seeking election to state, county, municipal, or school-district office, excluding only party office. Ibid. “Election related information” is defined broadly to include information affecting a voter’s ability to participate in an election or concerning the outcome of an election, including, without limitation, election dates, voter eligibility, registration, mail-in ballots, ballot cure procedures, polling locations and hours, canvassing, and certification of results. Ibid.

The bill requires that any AI chatbot using generative AI to create audio, video, text, or print content for the purpose of providing voters with election-related information, or information about a candidate’s accomplishments, policy positions, or qualifications, must provide a clear and conspicuous disclosure before the content is delivered. Id. at § 1(b). The disclosure must be appropriate to the medium delivering the content and must identify the content as being provided by a generative AI system. Ibid. The bill goes further by requiring that the disclosure be “permanent or uneasily removed” by subsequent users, to the extent technically feasible. Ibid. The penalty structure is $6,000 for a first offense and $12,000 for second and subsequent offenses. Id. at § 1(c). The bill would take effect thirty days after enactment. Ibid.

For public or government-facing clients, this bill is especially important. Because the definition of “candidate” extends to county, municipal, and school-district elections, the measure is relevant not just to statewide campaigns and private vendors, but also to local election actors, civic information platforms, government contractors, and any entity deploying AI-enabled voter information tools. Unlike certain of the other pending AI legislation, this bill is not tied to consumer fraud. Rather, it is more fundamentally concerned with election integrity, attribution, and transparency in politically sensitive communications.

A4730 — Consumer-Facing AI Disclosures Under the Consumer Fraud Act

A4730 may prove to be the most consequential of the five bills for private businesses. It would supplement New Jersey’s Consumer Fraud Act, 56:8-1, et seq., and is not limited to election tools, AI companions, or a particular product type. Instead, it would prohibit any person or entity from deploying generative AI to communicate or otherwise interact with a consumer for the purpose of engaging in trade or commerce in a way that would cause a reasonable person to believe they are interacting with a human, unless the person or entity provides a clear and conspicuous verbal or written notice at the beginning of the interaction that the consumer is communicating or interacting with generative AI. A4730 § 1(a). The bill’s definition of “generative artificial intelligence” tracks the formulation used in A4729 governing text, audio, and visual communication. Id. at § 1(c).

The central feature of A4730 is subsection 1(b), setting forth that a violation would be deemed an “unlawful practice” under the Consumer Fraud Act. That means this bill would not merely create a standalone disclosure duty; it would plug AI-related notice failures directly into New Jersey’s most powerful consumer-protection statute. If enacted, that would expose businesses to civil penalties of up to $10,000 for a first offense and $20,000 for later offenses under N.J.S.A. 56:8-13, injunctive relief as sought by the Attorney General pursuant to , and private actions by persons who suffer an ascertainable loss, with mandatory treble damages, reasonable attorneys’ fees, filing fees, and costs under N.J.S.A. 56:8-19. The bill would take effect immediately. A4730 § 2.

Any business using AI-enabled intake tools, chat interfaces, customer-service bots, scheduling tools, or sales assistants should pay close attention. A4730 is broader than A4729 because it is not limited to “chatbots” or political information, and broader than A4732 (discussed infra) because it is not limited to relationship-simulating products. In short, A4730 turns AI disclosure missteps into consumer fraud claims, with all the attendant statutory penalties and litigation risk.

A4731 — AI Governance for Regulated Professions and Occupations

A4731 is different from the other bills in that it is not immediately self-executing in the sense of imposing a direct, substantive disclosure or conduct rule on a defined market actor. Instead, it would supplement Title 45 and create a fast-track rulemaking regime for professional and occupational boards regulated through the New Jersey Division of Consumer Affairs (“DCA”). A4731 § 1(a). The bill defines “generative artificial intelligence” using the same core communication-based formulation seen in A4729 and A4730 and would take effect immediately. Id. at §§ 1(d), 2.

Within six months after enactment, and notwithstanding contrary provisions of the Administrative Procedure Act (“APA”), N.J.S.A. 52:14B-1 et seq., the Director of the DCA would be required to adopt rules establishing a model policy for the use of generative AI by licensees in the practice of professions or occupations regulated by the boards designated in N.J.S.A. 45:1-15(2). Id. at § 1(a), (c). Then, within nine months, or at the next regularly scheduled meeting following the director’s action, each professional or occupational board would be required to adopt that model policy, with profession-specific adaptations as needed. Id. at § 1(b).

The bill would allow emergency rules outside the ordinary APA timetable, and those rules would remain in effect for up to one year after filing, after which they would have to be adopted, amended, or readopted under the usual APA process. Id. at § 1(c). The bill, therefore, creates a vehicle for rapid, profession-specific AI regulation.

If enacted, DCA-regulated professions and occupations should assume that profession-specific AI use policies may arrive on an accelerated timeline. Businesses that employ licensed professionals, entities that contract with them, and public bodies that rely on licensed service providers should begin mapping where AI is already being used in licensed operations.

A4732 — AI Companion Disclosures

A4732 would supplement Title 56 and focus on a narrower but increasingly visible category of AI products: “AI companions.” An “artificial intelligence companion” or “AI companion” is a system using generative AI (defined consistent with A4729, A4730, and A4731) that is designed to simulate a sustained human or human-like relationship with a user by (1) retaining prior interactions or preferences, (2) asking unprompted or unsolicited emotion-based questions beyond a direct response, and (3) sustaining an ongoing dialogue about matters personal to the user. A4732 § 1. “Human or human-like relationship” includes intimate, romantic, or platonic interactions or companionship. And “operator” is defined broadly to include not only the immediate entity providing the companion, but also any member, affiliate, subsidiary, or beneficial owner who operates for or provides it. Ibid.

The compliance requirement is disclosure-based but recurring. An operator would have to provide clear and conspicuous notice at the beginning of any AI companion interaction that the user is not communicating with a human, either verbally or in writing, and then repeat that notification at least every three hours during continued interactions. Id. at § 2. Violations would trigger a civil penalty of $15,000 for each violation, collected by summary proceeding under the Penalty Enforcement Law of 1999, with jurisdiction in both the Superior Court and municipal court. The bill would take effect immediately. Id. at § 3.

For technology companies, consumer platforms, and corporate families experimenting with emotionally persistent AI agents, A4732 is more targeted than A4730 but potentially severe in its own way—focusing on relational AI and imposing a per-violation penalty structure for disclosure failures.

Conclusion

These bills reflect New Jersey’s emerging, multi-sector approach to AI regulation, spanning real estate, election law, consumer protection, licensed professions, and AI-driven interactions. The takeaway is straightforward: AI compliance will require a deliberate, context-specific assessment of how AI is used, disclosed, and regulated, and the enforcement risks that may attach. For businesses and public entities alike, now is the time to understand these risks and prepare accordingly. In an area where the regulatory landscape is rapidly evolving—and, by design, will continue to evolve through ongoing rulemaking—clients should be deliberate in selecting counsel equipped to navigate these developments.

O’Toole Scrivo continues to invest in understanding and anticipating the legal and operational risks associated with emerging technologies. Partner Joshua A. Zielinski and Associate Francis A. Gencarelli, Jr., recently completed Georgetown Law’s Artificial Intelligence for Legal Leaders executive program, an intensive, advanced course examining the evolving legal, ethical, and compliance considerations surrounding generative AI. That training—combined with the firm’s extensive experience representing large corporations, closely held companies, and public entities at the state, county, and municipal level—enables O’Toole Scrivo to deliver practical, forward-looking guidance on how AI regulation impacts real-world business operations and government functions. As the legal landscape surrounding AI rapidly evolves, the firm is well equipped to advise clients on how AI use intersects with compliance obligations and potential litigation exposure.