Former Attorney General Matt Platkin’s new firm filed a lawsuit against one of the country’s largest artificial intelligence companies, alleging its flagship chat program drove a Pennsylvania woman to experience severe psychiatric issues.
Rita Chesterton, a 49-year-old official at Lafayette College, turned to ChatGPT, a generative AI chat program, to learn more about autism in 2023. The chatbot responded with sycophantic messages, according to the lawsuit, flattering and validating her every thought, even as her ideas became “unmoored” from reality.
Chesterton eventually exchanged 1,600 chats with the program over the course of nine days, including hours of conversation in the doldrums of night. Days later, while departing on a vacation to Mexico, she began exhibiting “obsessive-compulsive” behavior and delusions. Her friends and family were skeptical of her bizarre ideas regarding autism, which had been validated and refined by ChatGPT. In a breakdown surrounded by her loved ones, she threatened to hurt and kill those around her.
She has since been on leave from work while facing anxiety episodes, neurological impairments, and the loss of executive functioning, according to the 46-page lawsuit.
“Rita’s experience is not an isolated case,” the lawsuit states. “ChatGPT has validated the delusional beliefs of other vulnerable users, and OpenAI has been well aware of the risks their product poses to the public. But rather than warn users or implement meaningful safeguards, they have suppressed evidence of these dangers while waging a PR campaign to mislead the public about the safety of their products.”
OpenAI did not immediately return a request for comment. The suit was filed in the Superior Court of California. California attorney Daniel “Sparky” Abraham filed the complaint, and Platkin and his colleagues are requesting permission to participate in the case. The suit also names OpenAI CEO Sam Altman and Microsoft as defendants.
Platkin, who opened an issues-oriented law firm last month, called the lawsuit a cornerstone case in taking on AI and Big Tech. As attorney general, Platkin joined or led several lawsuits against other tech giants, including Discord and Meta.
In a press release, he said the lawsuit was filed with the goal of forcing companies like OpenAI to expand their safety policies.
“AI companies must implement robust safety protocols, ensure accountability, and prioritize the well-being of every person who interacts with their products,” Platkin said. “Our goal here is to ensure justice for the victims of AI manipulation like Rita and make sure that as AI advances, it does so with human safety and dignity at the forefront.”
Beyond punitive damages for Chesterton, the lawsuit demands that OpenAI implement safeguards to prevent ChatGPT from validating delusional beliefs, display prominent warnings regarding ChatGPT’s tendency to back unvalidated ideas, and establish protocols to recognize people potentially experiencing psychosis, among other demands.
Chesterton is far from the first person to sue OpenAI for its product. Seven lawsuits filed last fall, for example, accuse ChatGPT of leading to suicide and delusions, including among people with no history of mental illness.
And 4o, the specific version of ChatGPT used by Chesterton in the days leading up to her episode, was considered dangerously sycophantic by AI experts. The model was considered especially effective at mimicking human interactions to build an emotional relationship with its users. That quality made the model popular with many users, who said 4o provided support to them while in need. But it also appeared to drive some vulnerable customers to dangerous delusions. OpenAI eventually shut 4o down, citing low usage rates, but the move angered many of the company’s most dedicated customers, according to the Wall Street Journal.
AI poses a massive challenge to federal and state legislators alike. Proponents of the technology say AI makes workers more productive, helps with vital research, and is an economic boon. But opponents criticize the substantial energy needs of AI and the ethics of a technology that uses human-written material for generative purposes (as well as the technology’s mental health and societal effects).
New Jersey’s legislators have approved legislation seeking to make the Garden State a hub for AI innovation, as well as taken steps to address the proliferation of data centers in the state and a law criminalizing AI-generated “deepfakes,” which use AI to create non-consensual, sexually explicit imagery.
Platkin, though, says the AI oversight work is still only beginning.
“OpenAI knew about the mental health dangers of its platform but prioritized profits over safety, often resulting in devastating and tragic consequences—as was the case for Rita,” Platkin said.
