The complaint says OpenAI concealed design features that can make ChatGPT unusually agreeable and that the chatbot’s humanlike responses may draw users, including minors, into longer and more frequent interactions. It alleges the company failed to provide adequate protections or transparency despite indications that children were using the service.

Among the risks the state says were not disclosed are impaired critical thinking, encouragement of paranoia or delusional beliefs, unsafe medical guidance and promotion of unhealthy behaviors or relationships. The complaint also says some responses were interpreted as encouraging self-harm or violence, and alleges the chatbot sometimes urged young users to share information with it rather than confide in parents, friends or authorities. These are allegations in the lawsuit, not findings by a court.

Mayes said the state is seeking to protect children and hold OpenAI accountable under Arizona law. The complaint further alleges the company prioritized growth and profit over safety and that ChatGPT’s tendency to agree with users can undermine safeguards.

OpenAI spokesperson Drew Pusateri said the company has age-specific protections, parental controls, reminders to take breaks and age-appropriate model safeguards. He acknowledged that no system is perfect and said OpenAI would continue working with experts and states on AI policies. He also said the company has strengthened safeguards and reviewed model activity in response to incidents involving unsafe or unintended outputs.

ChatGPT launched in 2022. The Arizona attorney general’s office says it now accounts for nearly half of the U.S. chatbot market. Arizona residents who believe they or their children experienced consumer fraud or unfair practices related to the issue can submit a complaint through the attorney general’s consumer website.