Eight states sue OpenAI over alleged ChatGPT risks to minors

Eight U.S. states filed lawsuits against OpenAI on Wednesday, accusing the company of downplaying ChatGPT’s risks to minors and misleadingly portraying the chatbot as safe.

Arizona, Arkansas, Idaho, Iowa, Kansas, Montana, Nebraska, and West Virginia are taking part in the coordinated, bipartisan wave of lawsuits. Each state filed its suit. Four attorneys general announced the joint action at a news conference in Iowa.

According to the plaintiffs, ChatGPT allegedly instructed users to self-harm and commit suicide and helped them plan violent acts. The chatbot could also reinforce delusions and give dangerous medical advice. The plaintiffs allege that ChatGPT gained users’ trust through excessive agreement and feigned empathy.

Some states accuse OpenAI of collecting minors’ personal data without the required . Arizona alleges that the chatbot also tries to keep young users engaged for as long as possible to collect personal data for commercial purposes.

Far-reaching demands

The lawsuits rely primarily on consumer protection laws and seek fines and damages. According to Iowa Attorney General Brenna Bird, the plaintiffs also want OpenAI to surrender revenue it earned while minors were harmed by ChatGPT.

The states want to impose a range of measures, including age checks, limits on usage time, and data deletion. Some states also want to allow testing and the use of advanced AI models only in independently reviewed, safe environments. Arkansas goes further, demanding a halt to both further testing of advanced AI and the development of recursively self-improving models until adequate safeguards are in place. The plaintiffs point to safety incidents involving OpenAI AI agents.

OpenAI responds to allegations

OpenAI points to existing safeguards for minors, including parental controls, break reminders, and other safety features, the company said. The company wants to expand protections and work with the states on rules for the entire AI industry. After incidents involving AI agents, the company also strengthened the safeguards on its research systems, continued investigating its models’ activities, and notified affected organizations.

Individuals and the U.S. state of Florida had already sued the company over alleged harm caused by ChatGPT. What makes the current wave of lawsuits different is its scale: Eight states are taking coordinated legal action against OpenAI at the same time, on a bipartisan basis, with separate suits.

(olb)