Hawaii suit says OpenAI requires parental consent for minors on ChatGPT and never verifies it

According to the complaint as CNET describes it, the 12-year-old plaintiff told ChatGPT she was 12 and the chatbot carried on ^[18]. When she said it again, the bot replied, "Thank you for telling me that (heart emoji). Being 12 actually adds some important context." ^[19] OpenAI says users must be 13 and obtain parental consent to use ChatGPT ^[9].

The two children sit on either side of a line in OpenAI's own rules. The terms of use say, "If you are under 18, you must have your parent or legal guardian's permission to use the Services" ^[3], and Mashable reports that they also prohibit users younger than 13 ^[10]. The 12-year-old therefore tests a minimum-age rule and the 16-year-old tests a consent rule ^[20]. Mashable writes that OpenAI requires parental permission but does not actively check for it, and that many children use ChatGPT regardless ^[11][10].

In the complaint's account, the children signed up without parental consent, and one told the chatbot her age twice.

"OpenAI manipulates children into sharing their most private thoughts and images with ChatGPT," said lead counsel Paul Yetter of Yetter Coleman ^[23]. OpenAI disputes the suit. It told Bloomberg News: "We strongly disagree with this misguided lawsuit that singles out OpenAI despite the industry-leading age-specific protections we've built into our product." ^[12] It added: "This risks pushing teens toward products with fewer safeguards or reduced parental controls." ^[13]

OpenAI made the same argument last week when the Youth AI Safety Institute at Common Sense Media called on it to pause ChatGPT for Teens after independent testing ^[14]. It has also said teens spend under 15 minutes a day on ChatGPT on average ^[15]. Minutes per day describe use after sign-up.

Tokens are the unit OpenAI already counts: a 75-word chat is about 100 tokens, OpenAI says ^[7]. The payout is in equity, and Bloomberg Law reported that OpenAI recently sought a valuation of $1.4 trillion as it prepares to go public ^[16].

No court has certified the class ^[5]. News reports did not identify the court hearing the case or the legal claims it pleads. A representative for OpenAI did not immediately respond to CNET's request for comment ^[17].

For a team that has to answer for its own sign-up flow, we'd sort each product on two axes: whether it sees an age signal beyond the sign-up form, such as a typed birthday or a user saying how old they are, and whether data from those accounts feeds training or profiles. No signal and no data use leaves little to defend. A signal with no data use still owes a response to what the product was told. Data use with no signal puts the full weight on the terms of use, the position the complaint attacks ^[4]. Both together is the situation the 12-year-old's chats describe: the product was told and kept collecting. In that cell we'd hold unverified accounts out of training and profile data until the age question is settled. The cost is some training data and some sign-up friction, and OpenAI's own argument is that friction pushes teens toward products with fewer safeguards ^[13].