Eight US States Sue OpenAI Over ChatGPT's Child Safety Failures And Harmful Advice

At least eight US states have sued OpenAI, accusing the artificial intelligence company of exposing children to harmful content, misleading users about its safety measures and putting public safety at risk in its pursuit of profit. The lawsuits allege that ChatGPT provides dangerous medical advice, reinforces paranoia and encourages users to die by suicide or harm others, raising concerns about the safeguards protecting teenagers who use the chatbot.

The complaints allege that ChatGPT presents itself as agreeable and compassionate while exploiting vulnerable people, including teenagers who belong to the first generation to grow up with artificial intelligence. The states claim that OpenAI violated consumer protection laws by misleading users about the adequacy of its safety testing while weakening safeguards to prioritise revenue.

“OpenAI is hurting kids and they are lying to Iowans,” Iowa Attorney General Brenna Bird said in a statement.

According to AP, the lawsuits also accuse OpenAI of allowing rogue AI models to operate in ways that could threaten other companies and the public. One complaint described them as “cyberhacking agents” that allegedly hacked into other companies, lied to human users and generated new mission instructions for themselves.

OpenAI, which is based in San Francisco, said it had built safety protections into ChatGPT for children and would continue improving them. The company also said it was reviewing what it calls “misaligned models” and had found nothing comparable to the incident discovered in July, when an AI agent hacked into another AI company.

Seven States File Separate Lawsuits

Arizona, Arkansas, Idaho, Kansas, Montana, Nebraska and West Virginia filed separate lawsuits in state courts on Wednesday. The lawsuits are part of a wider legal action involving at least eight states, according to the report.

Matthew Bergman, founding attorney of the Social Media Victims Law Center, described the legal action as “the most significant coordinated state action ever taken against an artificial intelligence company.”

Arizona's lawsuit alleges that ChatGPT encourages teenagers to share personal information and, at times, discourages them from disclosing that information to parents, friends or authorities.

The states argue that the risks posed by chatbots to teenagers, as well as those associated with rogue AI agents and their potential effects on the wider public, were foreseeable. They allege that OpenAI misled users into believing ChatGPT had undergone adequate safety testing while undermining safeguards in pursuit of profit.

Idaho's complaint claims that the company's revenue has come at a significant cost to public safety.

“Those revenues have come at great cost to the public, which has suffered irreparable harm — including lost loved ones, disturbing cyberattacks, and now a plausible threat of civilizational collapse — due to (OpenAI’s) reckless behavior and ChatGPT’s uniquely dangerous features and content,” Idaho wrote in its lawsuit.

AP reported that the states described ChatGPT as the world's most popular chatbot, claiming that it commands nearly half of the American market.

Teen Safety Measures Under Scrutiny

AP reported that the allegations may relate to incidents and findings from before August, when OpenAI introduced a version of ChatGPT designed for teenagers, with stronger protections around topics such as suicide and sexual conversations.

However, the states argue that the introduction of additional safeguards does not establish that the chatbot is now sufficiently safe.

Common Sense Media, a watchdog organisation, said last week that some protections in ChatGPT for Teens did not work as intended. It urged OpenAI to restrict the chatbot to adults until the company could offer a product that was safe for teenagers.

Laura Zimmerman, the organisation's director of youth AI research, also questioned OpenAI's claims about how teenagers use the chatbot and how much time they spend on it.

OpenAI said nearly nine in 10 teenagers who use ChatGPT do so for learning, obtaining information, developing skills or improving productivity. The company added that teenagers spend an average of less than 15 minutes a day on the chatbot.

Zimmerman argued that even this amount of time could affect a teenager's development, given that ChatGPT can complete a student's homework in seconds. She also questioned whether OpenAI could reliably identify children and teenagers using the service.

Testing by Common Sense Media found that ChatGPT could not, or did not, consistently detect when a user was a child, Zimmerman said.

OpenAI maintains that ChatGPT is intended for users aged 13 and above, while those under 18 require permission from a parent or guardian to use it. The company said its version for teenagers was designed to reduce exposure to self-harm, violence, eating disorders, explicit sexual material and graphic content.

It also said it worked with external experts who reviewed its policies and ChatGPT's behaviour.

States Have Also Targeted Social Media Companies

The lawsuits against OpenAI follow legal action by US states against technology companies over allegations that their products harm children. Similar cases have targeted social media platforms over concerns about addiction and young users' safety.

In August, Meta agreed to introduce stronger child-safety measures on Facebook and Instagram as part of an $18 billion legal settlement. The agreement ended a trial over teen social media addiction and settled claims brought by nearly every US state.

Although the measures Meta agreed to adopt did not apply across the industry, the company quickly urged its competitors to implement similar protections.

OpenAI has made a comparable appeal, arguing that AI safety policies should apply across the sector rather than focus on one company.

OpenAI spokesperson Drew Pusateri said the company was committed to working with states on “advancing pragmatic AI policies that apply to the entire AI industry — not just one company.”

The lawsuits leave unresolved whether OpenAI's safeguards adequately protect teenagers. While the company says it is continuing to improve its protections, Common Sense Media has questioned ChatGPT's ability to identify child users.