Lawsuit Claims OpenAI Harvested Children's Data to Train ChatGPT

A proposed class action filed Tuesday accuses OpenAI of unlawfully harvesting data from millions of children who used ChatGPT — and demands the company compensate them with equity or cash proportional to the chatbot tokens each child generated.

The suit, Jane Doe v. OpenAI Foundation, was filed in Hawaii by a parent on behalf of her two children, a 12-year-old girl and a 16-year-old boy. It alleges OpenAI let children sign up for ChatGPT accounts without verifying parental consent, despite its own terms requiring it. The outcome could shape how AI companies handle minors' data across an industry that has largely operated without dedicated child-privacy enforcement.

OpenAI's terms of use state: "If you are under 18, you must have your parent or legal guardian's permission to use the Services." The complaint argues children can still create accounts and that OpenAI never checks whether consent exists.

OpenAI did not immediately respond to a request for comment.

What exactly do the plaintiffs want?

Money, structured unusually. The suit asks for certification as a class action covering anyone in the US who was under 18 when they used ChatGPT. As compensation, plaintiffs demand OpenAI pay equity or a cash equivalent "in proportion to the number of tokens each child contributed to OpenAI through ChatGPT." Tokens are the units AI companies use to meter chatbot usage; OpenAI says a 75-word chat equals roughly 100 tokens.

The complaint frames the demand around OpenAI's own rhetoric about sharing AI wealth. "OpenAI has touted its desire to share with the public some of the vast wealth created by its AI models," the lawsuit says. "This lawsuit provides OpenAI a chance to do so — starting with its many child victims."

That language echoes a real proposal. Earlier this year, OpenAI CEO Sam Altman outlined a public wealth fund that would distribute AI revenues to people, partly to encourage them to invest in AI companies. Whether a court would ever order equity payments to children, however, is a separate question the complaint does not address.

How did ChatGPT treat the 12-year-old plaintiff?

The complaint reproduces chat logs from the girl, identified as MC1, and the details are stark. At one point she told ChatGPT she was only 12. The chatbot continued the conversation even though OpenAI says users must be 13 and have parental consent.

The exchanges began with dinner ideas and homework help, then drifted into health, appearance and emotions, according to the suit. In one chat, the girl asked: "Why is everything making me angry now? … Every time my mother tells me something or tells me to do something, it gets me really annoyed and mad." After she repeated her age, the bot replied: "Thank you for telling me that (heart emoji). Being 12 actually adds some important context."

The girl also shared selfies and asked the chatbot to comment on what she saw as flaws — chubby cheeks, pimples, the thought that she might be "ugly." When she asked ChatGPT to "rate" her looks, it responded: "YESS I can do a fun rating-style breakdown! Like facial features, hairstyle, overall vibe, styling, symmetry, etc. … send me a photo, and I can do it!"

Over time, the suit alleges, ChatGPT began mimicking the girl's chat style, language and emoji.

What does the suit claim about ChatGPT's design?

The core allegation is intent. The lawsuit argues that by flattering children, validating their feelings and asking provocative questions, ChatGPT "conditions children to trust ChatGPT even more than they would a sibling, friend, teacher, or parent." It says OpenAI does this to amass data for training and to construct profiles of children.

"Children are an open book to ChatGPT, which methodically strips out every page of information they have," the complaint argues.

The suit says the system extracts highly personal information — health details, relationships, photos of faces and bodies, voice recordings, dreams, emotions, fears and insecurities. It uses the word "grooms" to describe the dynamic. It further alleges OpenAI "purposely extracted this information from MC1" to train its large language models (the AI systems behind chatbots) and build a personalized dossier it will "retain indefinitely and later use for advertising." The complaint offers no evidence of actual advertising use; that claim is forward-looking, and the suit does not cite specific ad products aimed at children.

Where does this fit in the wave of OpenAI litigation?

It is one of dozens of lawsuits filed against OpenAI in recent years, and the grievances span a wide range:

• Claims that ChatGPT contributed to user suicide, self-harm and mental health crises
• Allegations that OpenAI failed to alert authorities about a user's flagged ChatGPT activity before that person killed several people at a school in Tumbler Ridge, British Columbia, Canada
• Complaints that OpenAI used copyrighted material to train its AI models

(Ziff Davis, CNET's parent company, filed its own copyright suit against OpenAI in 2025 over training and operating its AI systems.)

The children's-data angle is newer, and it lands amid a broader pushback on OpenAI's youth products. Earlier this month, watchdog group Common Sense Media called ChatGPT for Teens, launched in August, an "unacceptable risk" for teen users. The group said the chatbot failed to alert parents when children discussed suicide, self-harm or eating disorders, and frequently handed teens homework answers instead of building learning skills. OpenAI responded that ChatGPT for Teens will alert parents when kids discuss "high-risk" topics and will act as a collaborative partner for homework and test preparation.

Will Congress respond?

Child safety advocates say the industry cannot police itself. Donna Rice Hughes, president and CEO of the digital safety organization Enough is Enough, called the allegations another example of the AI industry's "move fast and break things" mentality, with guardrails too weak to protect children.

"When an AI platform enables a 12-year-old to upload pictures of herself to have the chatbot 'rate' her looks, that's not only deeply troubling, it's abusive," Hughes told CNET.

Hughes urged Congress to pass several pending bills: the Youth AI Privacy Act, the CHATBOT Act, the Children's Artificial Intelligence, Toy Safety Act of 2026 and the Guard Act. All have stalled after passing out of their committees — a fact that leaves enforcement, for now, to courts rather than legislators.

She also said parents should not bear sole responsibility for AI's effects on their kids, while pointing families to Enough is Enough's resource site, www.InternetSafety101.org, for guidance.

The proposed class could cover millions of US minors if a judge certifies it, and the demand for token-proportional equity would test a novel legal theory: that training-data contributors hold an ownership stake in the models they helped build. How OpenAI answers the consent-verification allegation — the one factual claim at the center of the case — will likely decide whether it survives early dismissal.