Washington, July 28 — A US lawmaker introduced legislation that would bar artificial intelligence chatbot companies from offering minors features designed to foster emotional dependence, compulsive use and addiction.
The Addictive Design Act, introduced by Democratic Representative Becca Balint of Vermont, would impose penalties on companies that make chatbots with addictive design features available to young users.
The legislation comes as teenagers increasingly turn to AI chatbots for mental health support. Most chatbots, however, are not designed to respond safely when a young person is experiencing a mental health crisis, according to material released by Balint’s office.
“Technology companies have spent years learning how to capture our attention and keep us engaged for as long as possible. They should not be allowed to apply those same tactics to vulnerable kids who may be turning to an AI chatbot during a mental health crisis,” Balint said.
“We can't make the same mistake with AI. We need clear rules, research, and accountability before more families are forced to suffer the consequences,” she added.
The bill would prohibit companies from allowing young people to access chatbots containing features known to increase addiction, unhealthy emotional attachment or compulsive use.
It identifies typing bubbles, avatars and the storage and use of personal information from previous conversations as potentially addictive features. The definition also covers requiring further engagement or payment to continue conversations, allowing conversations to last more than two hours and impersonating a real person.
Companies violating the proposed prohibition could face civil penalties of up to $5,000 for each affected minor.
The measure would establish an interagency task force to recommend ways to address the mental health risks associated with chatbot use among young people. It would also authorise $3 million for scientific research and $5 million for outreach programmes helping parents and educators identify warning signs.
Companies would be required to provide relevant data to researchers receiving grants under the programme. Refusing to provide the required data could result in penalties of up to $10 million.
The bill would also require companies using age-verification systems to minimise the information collected and delete it within 24 hours.
“AI chatbots should never be designed to keep young people emotionally dependent or endlessly engaged at the expense of their mental health,” said Rachel Madley, executive director of the Center for Health and Democracy.
Dr Marketa M. Wills, CEO and medical director of the American Psychiatric Association, said AI could improve mental health care but must be grounded in evidence. “AI chatbots should never replace the physician-patient relationship, which remains essential to safe, evidence-based, patient-centered care,” she said.
Representatives Chris Deluzio, Tom Suozzi and Bonnie Watson Coleman are co-sponsoring the legislation. Several mental health, social work and consumer advocacy organisations have endorsed it.