For teenagers around the world, including in India, AI is becoming an integral part of everyday life. They are using it to study for exams, comprehend difficult concepts, create content, and learn new skills. Global OpenAI data suggests that nine in ten teens already use ChatGPT for learning and skill building in a single week. Users in India are twice as likely to turn to ChatGPT for educational queries compared to the global median.

The question, then, is how that usage will shape the way they learn, think, and navigate the world.

AI has the potential to personalise learning, bridge educational gaps, and make knowledge more accessible through native-language and voice experiences. But greater access entails greater responsibility. Teens need protection from harmful and age-inappropriate content, and from patterns of use that could weaken independent thinking or encourage over-reliance.

This is why teen safety must be about more than simply restricting content. AI products should also help teenagers develop the skills and judgement to use them safely, critically, and independently. That requires building upon safeguards already designed into AI systems with additional protections for teens. This means thinking about age-appropriate content and parental controls, but also how the system responds, what defaults it sets, when it slows interactions down, and how it reinforces real-world support.

In India, age-appropriate design must also reflect how teens actually access technology. In many Indian households, access happens on shared devices, with some surveys suggesting that such usage for teens may be well over 70%. Personal device ownership is likely even lower among girls than boys. These nuances need to be accounted for in how protections are designed.

On shared devices, an age declared or verified when an account is created may not always reflect the person actually using an AI service. Age-prediction mechanisms can then provide an additional layer of protection by using a range of signals to assess whether a user is under 18 and adapting the experience accordingly. ChatGPT relies on such age estimation, using behavioural and account-level signals to identify under-18 users and activating additional protections automatically.

But AI companies cannot solve teen safety alone. Parents, educators, and trusted adults have an important role in helping teens build agency. This means knowing AI can be wrong, recognising unhealthy patterns of use, and stepping away when needed.

That shared responsibility creates a second challenge: helping the adults around teenagers keep pace with what constitutes responsible use. Parents and educators are being asked to guide a generation growing up with a technology many are only beginning to understand themselves. That makes teen safety not just a product question, but a societal one. Good design should not force a trade-off between oversight and independence; it should help families navigate with more confidence and less guesswork.

For teens, safety is also about what AI encourages them to do. Does it nudge them to develop judgement, curiosity, and independent thought? For younger users in particular, enabling meaningful learning should be central to how AI products are designed.

That means prioritising learning over speed and convenience. While products for adults may prioritise quick answers and productivity, AI for teens should help them ask better questions and work through problems. The aim should be to make AI a tool that strengthens learning and independence, rather than one that simply makes it easier to avoid the work of learning.

These principles should inform how AI products are designed in practice. For example, parents can be given tools to shape how their teens use AI, including settings that make new chats default to a learning-oriented mode during designated study hours, guiding teens through problems step by step rather than simply providing answers.

New uses of AI will create new risks, and our understanding of how teens interact with these systems will continue to evolve. Companies need to keep testing and improving their systems. India now has a chance not just to adopt AI, but to help define global norms for what responsible AI should look like. Given its scale, diversity, and distinct access patterns, lessons from the Indian experience have relevance well beyond its borders.

If AI is going to grow up alongside this generation, we should make sure that this generation can grow with it safely.

Misra is Head of Strategy & Government Affairs, OpenAI India