Become an Independent member to bookmark this article

Want to bookmark your favourite articles and stories to read or reference later? Start your Independent Membership today.

Join today

Already a member? Log in

OpenAI is accused of exploiting children by using their private conversations with ChatGPT to train its AI models.

A proposed class action lawsuit by a mother of two in California accuses the AI company of using highly sensitive data from millions of children without consent.

“Through its artificial intelligence chatbot, ChatGPT, OpenAI has profited from our children on an unprecedented scale,” the 81-page complaint states.

“For years, it wrongfully has extracted vast amounts of highly sensitive and valuable data from millions of children.

“It then used that priceless trove of children’s information to train large language models, build detailed profiles of children, and enrich itself. All of this was illegal.”

The anonymous mother says her 12-year-old and 16-year-old children signed up to ChatGPT without parental permission this year.

OpenAI prohibits children under 13 from creating or using a ChatGPT account, while 13- to 17-year-olds require consent from a parent or legal guardian.

Despite the requirements, it is relatively easy to sign up to a ChatGPT account so long as a user has access to an email address and phone number.

A 2026 report by Common Sense Media found that “AI use is common” among pre-teens and teens.

open image in gallery

OpenAI is facing separate legal action brought about by eight US states that claim ChatGPT harms young users.

The lawsuits, filed on Wednesday, allege that OpenAI’s chatbot has encouraged dangerous behaviour, such as coaching teenagers to harm themselves or others.

Attorney generals from the eight states – Arizona, Arkansas, Idaho, Iowa, Kansas, Montana, Nebraska and West Virginia – share a common accusation that ChatGPT was designed to be highly agreeable and friendly in order to gain users’ trust.

OpenAI has previously acknowledged that one of its updates to ChatGPT caused it to be excessively sycophantic.

In April 2025, the AI firm rolled back its GPT-4o model that powers ChatGPT’s responses after admitting that it was “overly flattering or agreeable” to users.

“ChatGPT’s default personality deeply affects the way you experience and trust it,” OpenAI said at the time.

“Sycophantic interactions can be uncomfortable, unsettling, and cause distress. We fell short and are working on getting it right.”

The Independent has reached out to OpenAI for comment on both lawsuits, though the company does not typically comment on ongoing legal matters.

Join our commenting forum

Join thought-provoking conversations, follow other Independent readers and see their replies