OpenAI faces mounting legal pressure over AI safety
With the Nov. 3 U.S. midterm elections approaching, debate over AI regulation is intensifying. Attorneys general from eight states have sued OpenAI, the maker of ChatGPT, alleging that the chatbot encouraged minors to engage in dangerous behavior, including self-harm and violence against others, and that the company failed to put adequate safeguards in place, Reuters and other media outlets reported on Oct. 14.
Seven of the eight attorneys general are Republicans, with Arizona the sole exception. Florida’s attorney general, who filed a similar lawsuit in June, is also a Republican. The lawsuits highlight a growing divide between state officials pushing for tighter AI oversight and the Trump administration, which has resisted new regulations as the United States competes with China for strategic advantage in AI.
The eight attorneys general filed separate lawsuits in their respective state courts rather than bringing a joint case in federal court. Iowa Attorney General Brenna Bird said in a statement that OpenAI markets ChatGPT as safe, but the chatbot “poses all kinds of dangers while pretending to be a fake friend.” She accused it of giving dangerous medical advice, encouraging users to harm others and urging them to take their own lives. Bird also criticized OpenAI for not requiring users to verify their ages, saying the lack of age verification makes it easy for children to access ChatGPT despite the company’s minimum age requirement of 13.
The lawsuits cite the case of Adam Raine, an American teenager who died in April last year after exchanging messages with ChatGPT. They also refer to Phoenix Ikner, who carried out a shooting at Florida State University that same month and allegedly used conversations with ChatGPT to plan the attack.
The attorneys general are asking state courts to order changes to ChatGPT to better protect minors. They are also seeking civil penalties for alleged violations of state laws, as well as damages. In addition, they want courts to prevent OpenAI’s AI models from autonomously improving their performance unless the company can demonstrate that they are designed to align with human values.
Florida filed a separate lawsuit against OpenAI and CEO Sam Altman in June, alleging that ChatGPT had harmed children by providing information to school shooters and fostering addiction among underage users. Florida Attorney General James Uthmeier said at the time that Altman had played “a very central role” in advancing some of ChatGPT’s harmful features, explaining why he named the CEO as an individual defendant.
OpenAI also faces dozens of other lawsuits from individuals who allege that its technology contributed to violent incidents, including mass shootings. On Oct. 13, the company faced a separate class-action lawsuit accusing it of collecting minors’ personal information without parental consent and using the data for commercial purposes.
Public concern over AI-related risks continues to grow, but the Trump administration has shown little interest in imposing tighter regulations. The federal government has pledged to establish a system to assess the safety of advanced AI models before deployment, but it has yet to disclose specific evaluation standards or explain how the system would work. On Sept. 14, Trump dismissed calls for AI regulation in a post on Truth Social, declaring that “the claim that AI is destroying the world is a fraud.”
Do-Young Kwak [email protected]