OpenAI Locks Down ChatGPT for Under-18s as Teen Use Surges

The News

OpenAI has begun switching on a set of automatic, age-appropriate safeguards for ChatGPT users it estimates to be under 18, paired with a new suite of parental controls. The move, detailed in a company post, reframes the chatbot from a general-purpose tool into something closer to a supervised study companion for younger users.

The scale numbers explain the urgency. OpenAI says nearly nine in ten teenagers on ChatGPT use it every week for learning, information, skill-building or productivity. Some 18 million weekly users now tap its interactive maths and science experiences, which have expanded to more than 300 topics, while a pronunciation feature supports over 61 languages.

For accounts estimated to belong to under-18s, protections now activate by default. These include tighter content safeguards against graphic violence, self-harm, risky challenges, body-image material and inappropriate roleplay, plus more frequent break reminders during long sessions. A dedicated Study Mode, built with teachers and learning scientists, walks pupils through problems step by step rather than handing over finished answers.

Parents get their own dashboard. They can link a teen account, keep Study Mode on by default, set quiet hours, toggle voice mode, and restrict image generation. Crucially, the system can notify a parent when it detects signals of potential self-harm, and flag account deactivations tied to violent-threat policy breaches. OpenAI says it built these features with input from online-violence prevention group Moonshot and the Family Online Safety Institute, which it joined this year.

Why It Matters

The announcement lands as teen safety becomes the defining reputational battleground for consumer AI. The template is familiar from social media: the last time a technology this sticky reached this many young users at once, it was Instagram in the early 2020s, and the fallout from internal research on teen mental health reshaped how regulators viewed the entire sector. AI companies have watched that playbook and are trying to get ahead of it rather than defend after the fact.

There is commercial logic too. Age-gating is expensive to retrofit and easy to get wrong, and the firm that sets the default standard shapes what regulators later demand of everyone else. By publishing detailed controls now, OpenAI is effectively drafting the compliance template its rivals will be measured against.

Indian Angle

For India, the timing is pointed. The Digital Personal Data Protection Act, 2023 treats anyone under 18 as a child and requires verifiable parental consent before their data is processed, alongside a bar on tracking and targeted advertising aimed at minors. The draft rules circulated for consultation put operational detail on that promise, so OpenAI's parent-linked accounts and consent-driven design map neatly onto what Indian law will expect. That makes this less a courtesy and more a preview of Indian compliance.

India is also one of ChatGPT's largest and youngest user bases, with school and college students a core cohort. That collides with a domestic edtech sector still rebuilding trust after the BYJU'S implosion, and with home-grown model builders such as Sarvam and Krutrim that will now face a higher published bar for youth safety. For Indian parents and schools weighing AI in classrooms, a credible under-18 mode is the difference between adoption and a ban.

The regulatory read is that MeitY now has a live reference point. When Indian rule-makers debate what "age-appropriate design" should mean in practice, they can point to a shipping product rather than an abstract principle.

FAQ

Who counts as a teen under these rules?

OpenAI applies the protections automatically to any account it estimates belongs to a user under 18. Age-appropriate safeguards and Study Mode switch on by default, and parents can link the account to manage settings directly.

What can Indian parents actually control?

Linked parents can keep Study Mode on, set quiet hours, disable voice mode, and restrict image generation. They also receive alerts if the system flags potential self-harm signals, giving guardians oversight that aligns with the consent duties in India's DPDP Act.

How does this affect Indian edtech and model builders?

It raises the published bar. Domestic players such as Sarvam and Krutrim, and edtech firms courting schools, will be measured against a documented under-18 safety standard when pitching to cautious Indian parents and institutions.

Where can I read the original announcement?

OpenAI has published the full detail on its official company blog. The link appears in the attribution paragraph below.

This story was reported by OpenAI. Read the full original coverage at OpenAI.