The New Jersey Assembly’s Science, Innovation & Technology Committee will hear three AI disclosure bills on Monday, Oct. 19, moving a package that reaches directly into how small businesses market, sell, and talk to customers online.

The three measures on the agenda, all sponsored by Assemblyman Andrew Macurdy (D-21) with co-sponsors Park and Singh on the lead bill, approach AI from different angles. A5089, the AI Image Disclosure Act, would require both manifest and latent disclosures on any AI-generated image, video, or audio, with a $5,000 civil penalty per violation enforced by the Attorney General, and each day treated as a separate violation. A5090 would impose civil penalties on generative AI platforms engaged in harmful activity. A5275 reaches further up the stack, requiring large frontier developers, those above 10^26 training FLOPs and $100 million in annual revenue, to file Risk Management and New Model Risk Disclosures mapped to NIST’s AI Risk Management Framework.

The committee has already advanced a fourth bill, A4730, which the National Law Review says “may prove to be the most consequential” because it plugs undisclosed AI-to-consumer interactions directly into New Jersey’s Consumer Fraud Act. That framing matters. A chatbot handling product questions, an AI-written sales email, a generated product photo, all become CFA exposure if deployed without a clear, conspicuous notice.

For a 5-to-30-person business running AI-assisted acquisition into New Jersey, the compliance surface is now federated across labeling (A5089) and interaction disclosure (A4730). It’s the same regional pattern visible in New York City’s 10-bill AI package and in Connecticut’s CART Act, now law.

Monday’s vote will tell us which of the three advance. The Legislature’s session leaves roughly a two-to-three-week window of real movement.

Sources

Sources