OpenAI has launched ChatGPT for Teens, a version of the chatbot with stronger built-in safety protections for users under 18, the company said in a blog post on Monday. Teen accounts become the default for anyone the system estimates to be underage, or who states that they are between 13 and 17 during sign-up. Users under 13 remain barred from the platform. The default matters in practice: until now, age gates on ChatGPT, as on most platforms, have depended on users self-declaring a birthdate, an approach UK regulator Ofcom has said does not on its own meet the “highly effective” age assurance standard in its children’s safety rules.

According to the BBC, part of the rollout includes removing the chatbot’s ability to respond to teens using a human-like voice, a move intended to reduce the tool’s tendency to appear person-like.

OpenAI’s own announcement does not specifically describe that change, but it says teens will receive customisation options such as accent colours and voice variations designed to feel personal “without blurring the line between a useful tool and a human relationship.”

The company said the update is not a response to any single incident, but instead reflects broader industry concern about people forming strong attachments to AI systems and treating them as sentient. That concern has a legal backdrop elsewhere in the sector: companion-chatbot maker Character.AI has faced lawsuits from families over minors’ interactions with its service and has since introduced a separate model for teenage users with stricter guardrails.

OpenAI said its under-18 model rules now go beyond blocking romantic or sexualised roleplay. The chatbot should not use romantic language, encourage emotional dependence on it, or claim to have feelings.

ChatGPT study tools and parental alerts

The teen experience builds on Study Mode, which OpenAI says uses guided questions rather than direct answers to help students work through material. New “responsible homework reminders” are designed to detect when a teen is trying to shortcut an assignment and redirect them back into that step-by-step format. OpenAI is also adding quizzes, learning visualisations, and scheduled “Study Hours” that parents or teens can set for when Study Mode stays on by default, alongside “quiet times” that switch the tool off entirely.

On the safety side, OpenAI is adding eating disorder-related prompts to the list of flagged requests that trigger alerts to parents with linked accounts. The alerts extend the parental controls OpenAI rolled out for 13-to-17-year-olds earlier in 2025, which let parents link a teen’s account through Google’s Family Link and see how much time their child spends in the app. Each alert is reviewed by a person before being sent, and the company says it aims to reach parents within an hour.

OpenAI also announced a partnership with CodeAI, an education-focused AI company, aimed at helping teens and teachers understand how AI systems work, question their outputs, and use them responsibly in the classroom.

The company is not alone in tightening rules for younger users. Anthropic’s Claude remains restricted to adults only, and Google’s Gemini, launched as Bard, also excluded children at launch, though Google later opened Gemini to teenagers with its own restrictions. There is currently no dedicated AI regulation covering minors in the UK, though past government proposals considered restricting chatbot use by under-18s. The closest existing lever is the Online Safety Act, under which Ofcom’s children’s safety codes took effect in July 2025 with duties on in-scope services likely to be accessed by children; the regulator has separately begun examining how the Act applies to AI chatbots built into search services.