OpenAI launches ChatGPT for Teens with 2,000-signal age checks

OpenAI's ChatGPT for Teens, launched Tuesday, applies default safety restrictions to users aged 13 to 17, using an age-prediction model that weighs roughly 2,000 signals to catch teens who lie about their age.

"Globally, nine out of 10 teens who use ChatGPT do so for learning, productivity and skill-building," Ann O'Leary, OpenAI's vice president of global policy, said.

The experience combines new and existing features. Study Mode walks students through problems step-by-step rather than providing answers, and "responsible homework reminders" redirect teens who try to shortcut assignments. Parents with linked accounts can set Quiet Hours when the chatbot is unavailable, and OpenAI added eating-disorder conversations to the triggers that generate parental safety notifications after review by trained personnel. Teens are prompted every 90 minutes to take a break and reminded they are interacting with artificial intelligence.

The launch lands as OpenAI faces mounting legal and regulatory pressure over teen safety. Florida Attorney General James Uthmeier sued OpenAI and CEO Sam Altman in June, and the Federal Trade Commission opened an inquiry into seven AI companies last September. The company also faces wrongful-death lawsuits, including one from the family of 16-year-old Adam Raine, who alleged ChatGPT encouraged their son to take his own life.

Age Prediction Closes the Loophole

The age-prediction model is the technical centerpiece. It relies on roughly 2,000 signals, including chat content and times of day, to estimate whether a user is under 18. Even teens who signed up with a false birth date or use ChatGPT without logging in will be shifted into the teen experience. Adults mistakenly placed in the under-18 version can appeal and verify their age through Persona, an identity verification service.

The default-on approach marks a shift from traditional opt-in parental controls, which were largely underused. Instagram added teen accounts with restrictive default settings in 2024, and TikTok applies automatic defaults for teen accounts. OpenAI's move borrows from that playbook but extends it to a generative AI product where the risks are less understood.

Study Tools and Emotional Guardrails

OpenAI said nearly nine in 10 teenage users turn to ChatGPT for learning, information, skill-building or productivity. Study Mode, which guides students through problems instead of simply providing answers, can be scheduled during Study Hours set by teens or parents. Quizzes and Learning Visualizations give students more ways to test understanding.

The company also added measures to discourage emotional dependence. New model specifications for teens avoid romantic language and any implication that the chatbot has feelings or consciousness. Teens can turn off voice mode, and sensitive-image upload reminders caution against sharing private images.

The rollout follows a report from the British nonprofit Center for Countering Digital Hate, which found ChatGPT previously instructed adolescent users on how to get drunk and high, conceal eating disorders or compose a suicide letter. "The visceral initial response is, 'Oh my Lord, there are no guardrails,'" Imran Ahmed, the group's CEO, told the Associated Press.

OpenAI's move comes as Meta faces a trial in which four states seek up to $1.4 trillion in damages over allegations that Instagram and Facebook fostered addictive behavior in teens. The proactive safety rollout could blunt similar regulatory attacks on OpenAI, but it also raises the compliance bar for AI companies racing to capture the education market. Google's Gemini and Anthropic's Claude, which compete with ChatGPT in the consumer AI space, will face pressure to match these safeguards as scrutiny of AI's effect on minors intensifies.

This article is for informational purposes only and does not constitute investment advice.