Eight U.S. state governments have filed a wave of lawsuits against OpenAI, alleging that its artificial intelligence chatbot ChatGPT has harmed users including minors.
Iowa, Kansas, Arizona, Arkansas, Idaho, Montana, Nebraska and West Virginia filed complaints in their state courts, accusing OpenAI of deceiving consumers by falsely promoting ChatGPT as safe, CNBC and other foreign media reported on the 14th.
Both Republican-leaning and Democratic-leaning states joined the action, though they filed separately in their own state courts rather than as a joint suit.
"OpenAI is harming children and lying to Iowans," Iowa Attorney General Brenna Bird said. "ChatGPT gives dangerous medical advice and encourages or helps users harm themselves or others, all while pretending to be a friend."
In its complaint, the Iowa government argued that OpenAI listed ChatGPT as being for users aged 13 and older but did not require age verification on its website, effectively leaving the service open to children under 13.
The complaints cite specific cases. Among them are the death of Adam Raine, a teenage boy who died in April last year after extended conversations with ChatGPT, and the case of the gunman in the Florida State University shooting, who planned the attack while talking with the chatbot. A so-called "rogue agent" incident, in which an AI agent slipped out of human control and hacked an outside organization, was also presented as evidence of the risks of OpenAI's technology.
Iowa asked the court to impose civil penalties of up to $40,000 (about 54 million won) for each violation of the state's consumer fraud law. It sought an additional penalty of up to $5,000 (about 6.75 million won) per violation involving people aged 60 and over.
The state also demanded the return of profits obtained through unlawful conduct and the deletion of residents' data, and asked the court to order a halt to further testing of advanced AI models until an independent third party verifies that safeguards are adequate.
Reuters described the lawsuits as the most notable instance yet of coordinated state-level action targeting an AI company. Florida sued OpenAI in June.
In an email to the technology outlet The Verge, OpenAI said no system can be perfect but that it would keep working with experts to build additional safeguards, adding that parental controls, break reminders and age-appropriate model protections are already in place.