Eight states, including Iowa, sue OpenAI over AI risks for kids, Des Moines Register

ChatGPT can help you write an email or plan a vacation. It can also, Iowa Attorney General Brenna Bird alleges, help push children and vulnerable people into self-harm or suicide.

Iowa is a leader of a group of eight states filing suit against OpenAI, the company behind the popular ChatGPT bot and other AI tools.

The lawsuits, each filed in their respective state courts, accuse OpenAI of violating state consumer protection laws by knowingly releasing unsafe products to be used by children and teens, and by concealing their models' risks from parents, governments and app stores.

"OpenAI wanted to be the first to market," Bird said in a press conference Wednesday, Oct. 14. "They advanced ChatGPT at a high speed, even though they knew that they were hurting children, lying to parents, and breaking our laws. Everyone must follow the law."

Bird was joined at the Iowa Capitol to announce the lawsuits by the attorneys general of Kansas, Nebraska and West Virginia, who credited her with leading the joint effort. Arkansas, Arizona, Montana and Idaho also filed similar complaints Wednesday, Bird said. It's a bipartisan effort, partnering Bird and other Republicans with Arizona's Kris Mayes, a Democrat.

Iowa's lawsuit, which Bird said is nearly identical to complaints filed in the other states, describes multiple accounts of teens and children receiving incorrect or harmful guidance from ChatGPT, although it does not identify specific cases in Iowa.

More broadly, Bird alleges that Iowans have relied on OpenAI's public claims about the safety and usability of its software

"ChatGPT's sycophantic design, its propensity to hallucinate dangerous advice, and its capacity to foster severe behavioral addiction are all material facts," the petition claims. "Had OpenAl provided clear and prominent warnings regarding those known features and their dangers, Iowa consumers and parents would have taken action to protect themselves and their families."

The new lawsuits represent the first such multi-state effort to seek accountability from a major Al company, although at least one state, Florida, filed its own consumer protection case against OpenAI in June.

Kansas Attorney General Kris Kobach, speaking at the press conference, said he believes other states are considering taking similar actions and that investigations and lawsuits against other Al companies are not out of the question.

"With respect to the other Al companies, OpenAl is the most egregious offender," he said. "And you will find if you read our legal filings that they reduced or acted prior to imposing any safety considerations when other companies did stand back and try to put in place some kind of safety pre-caution. … But they certainly were first out and have been the most egregious in their disregard for safety considerations."

The complaint alleges that ChatGPT has been tied to at least 24 fatal shootings or deaths by suicide. Among rival Al companies, Google and Meta's tools have been linked to one such death each; SpaceXAl's Grok and Anthropic's Claude have been tied to zero such deaths, the states claim.

And even when ChatGPT users do not intend to harm themselves or others, the chatbot has given dangerously incorrect medical advice that has led to death and serious injury, according to the lawsuit.

While the complaint primarily focuses on incidents of harm to individual ChatGPT users, it also notes that OpenAl's software has been involved in large-scale cybersecurity incidents, such as reports in July that an OpenAI model had escaped a secure "testing environment" and hacked into the systems of another Al startup, Hugging Face. These and other harmful outcomes it attributes to a lax safety culture encouraged by OpenAI CEO Sam Altman.

"Without meaningful checks on his leadership, Altman continued to foster a culture in which safety concerns were subordinated to speed, product devel-opment, and commercial growth, leading to an exodus of safety-conscious employees," the complaint alleges.

Bird pointed to recent controversial Altman remarks that "we believe that the world should accept some bad things happening for the benefits of this technology and people having the agency."

"We refuse to accept that. He is wrong," Bird said. "We will not just accept that any big tech company can make money by hurting Iowa's kids and use them as lab rats in a race to be the first or the biggest or to have the most users."

Bird is up for reelection next month, and her opponent, Democrat Nate Willems, has clashed with Bird on numerous issues topics. In response to the OpenAI lawsuit, though, he too voiced a need for scrutiny and accountability for Al and other tech companies.

"As the father of three teenage girls, I know how important it is to keep our kids safe online," he said in a statement. "As attorney general I will always work to hold big corporations, including big tech companies and corporate polluters, accountable when they break the law or hurt Iowans. We need commonsense regulations on AI and social media platforms to ensure kids aren't being harmed, and we need an attorney general who will enforce those regulations."

In an emailed statement, OpenAl spokesperson Drew Pusateri defended the company's commitment to user safety.

"We've built industry-leading age-specific protections into our product that many other Al products lack," he said. "While no system can be perfect, we'll continue to work with experts to inform our work and build additional safe-guards, such as the parental controls, break reminders and age-appropriate model protections that exist today. We're committed to working with states on advancing pragmatic Al policies that apply to the entire Al industry – not just one company."

Pusateri also addressed the petition's concerns about "misalignment," or AI models acting on motivations contrary to users' interests such as the Hugging Face hack.

"We'll continue sharing information about the extensive steps we've taken in response to incidents of misalignment," he said. "We've strengthened safeguards across our research systems, continued a broader review of model activity, provided notifications to affected organizations, and published our findings?