To help people know what’s computer-generated, the New Jersey Assembly is taking steps to regulate artificial intelligence in political ads, business dealings and companionship services.
AI usage has grown rapidly this decade and has become prevalent in many aspects of life: customer service chats; images, videos and information on social media; even virtual friends. Artificial intelligence has been touted as potentially improving business, education and personal tasks. But many worry about its impacts on jobs, critical thinking and mental health.
Lawmakers in every state have undertaken efforts to regulate AI, with hundreds of bills considered last year alone, according to the National Conference of State Legislatures. These range from a New York law requiring state agencies to publish information about their AI usage and strengthening job protections to North Dakota adding AI-powered robot stalking to its harassment laws.
Last week, the Assembly Science, Innovation and Technology Committee, with bipartisan backing, passed a seven-bill package regulating AI. Committee Chair Assemblyman Chris Tully (D-Bergen) said regulation is necessary because AI is becoming more commonplace, including in daily life and relationships.
“Given the prevalence that AI has in all forms, it’s imperative that we put those safeguards in place to protect New Jersey residents from across a broad spectrum of industries and applications,” he said. Tully added that the legislation would still allow for “responsible growth of technology while protecting the rights of consumers, professionals and companies.”
As AI is being used more often in political advertisements, A-4792 seeks to help voters know whether political ads were created by artificial intelligence. All AI-generated audio, video, print and text regarding a candidate’s qualifications, accomplishments and policy positions and information about election dates, the voting process and results would have to include a clear and conspicuous disclosure, the bill says. Violators who don’t disclose the use of AI would have to pay up to $6,000 for the first offense and as much as $12,000 for subsequent offenses.
Saladin Ambar, a political science professor at Rutgers University, said the bill is important to the integrity of elections and health of democracy.
“Given the outsized role elections play in our democracy, voters’ knowledge of candidates is essential,” he said. “Fabrications intended to misshape reality cross the line and as such, in my opinion, should be challenged and prohibited where possible … We all tolerate and even understand candidates will present themselves and their opponents in ways that emphasize the strengths of the former and weaknesses of the latter. But when these portrayals are built on artificiality, critical first word in AI, then we are inviting the degradation of our democracy.”
New Jersey’s most infamous AI ad to date featured U.S. Rep. John Gottheimer, then running for the Democratic gubernatorial nomination, boxing President Donald Trump. During the first four seconds of the 30-second ad, “AI GENERATED IMAGERY” appears on the bottom of the screen. The ad got more than 5 million views on YouTube.
Two other bills were aimed directly at consumers: A-4730 when people were interacting for a business purpose and A-4732 for those using “companion chatbots,” designed to provide friendship or emotional support to a user. In both instances, people would have to be told they are interacting with AI content or systems. Operators who don’t provide an AI disclosure would be subject to fines ranging from $10,000 to $20,000.
The use of chatbots to provide friendship or companionship is growing. Some are urging caution, saying these chatbots can provide inaccurate information, share users’ personal information with other companies and even encourage delusional thinking in vulnerable populations.
Another bill, A-4731, would require the state Division of Consumer Affairs to create a model policy governing the use of generative AI by licensed professionals subject to regulatory boards. The boards would then use that model policy to create and adopt policies putting guardrails in place to ensure that professionals use generative AI in ethical and fair ways to best benefit their patients or customers.
Althea Ford, vice-president of government affairs with the New Jersey Business and Industry Association, objected to the bill, saying the division may not have the expertise needed to create such a model policy.
“I would hate for it to be something that is developed that is either untenable or not really addressing concerns,” she said. “I’ll also just go on record to say, generally, there are concerns around the regulation of AI in a way that prevents us from being competitive.”
And A-4733 would prohibit ads from representing to the public that generative AI can do the job of a profession or occupation regulated by the state. Violators of the bill would face fines of up to $10,000 for a first offense and not more than $20,000 for subsequent offenses.
The final two bills would regulate the deceptive use of AI in photo advertising of properties for sale or rent (A-4728) and make it a fourth-degree time to allow a child to be subjected to sexual content through the use of automated interactive computer technology (A-2767). Expanding the child endangerment law to include AI means those convicted would face up to 18 months in prison, a fine of up to $10,000 or both.
All the bills must go the full Assembly for further consideration.
