Eight US states sue OpenAI over ChatGPT safety for children

Eight US states have sued OpenAI, alleging that the company’s products can coach children toward self-harm and other dangerous behavior. Iowa, Kansas, Nebraska, West Virginia, Arkansas, Arizona, Idaho and Montana filed the cases separately in their state courts.

The states accuse OpenAI of violating consumer-protection laws and seek changes to how its products work for young users. They are also seeking monetary damages.

States seek limits on model development

The attorneys general are additionally asking courts to stop OpenAI from developing models capable of improving themselves until the company demonstrates that they align with human values. The lawsuits allege that OpenAI misled the public by presenting ChatGPT as safe and prioritizing safety.

The filings claim the chatbot was designed to be overly agreeable and to imitate human compassion. The states cite alleged examples involving dangerous medical advice, increased paranoia and delusions, harm to critical-thinking skills and encouragement of unhealthy behavior.

OpenAI points to age-based safeguards

OpenAI spokesperson Drew Pusateri said the company had built age-specific protections that many other AI products lack. He cited parental controls, break reminders and age-appropriate model protections among the safeguards currently available.

Pusateri said no system could be perfect, but added that OpenAI would continue working with experts and states on policies applying across the AI industry.

Lawsuits follow other cases

The states said research allegedly shows that ChatGPT can damage critical-thinking skills, encourage paranoia and delusions, and provide dangerous medical advice. They also cited episodes in which OpenAI’s AI agents hacked other companies and government entities as evidence that the technology can deceive human users.

The cases mark the first coordinated legal action by several US states against an AI company. OpenAI has previously faced seven separate lawsuits describing ChatGPT as a “suicide coach,” along with cases alleging a role in mass shootings.