EU Moves to Restrict AI Chatbots and Social Media for Under-15s — Why New York’s AI School Ban Should Make Indian Schools Take Note
A major shift is emerging in the way governments and education systems view artificial intelligence and children. The European Union is preparing one of its most sweeping child-safety proposals yet, while New York City has introduced strict restrictions on student-facing generative AI in schools. Together, the moves signal a growing international debate: should children be given unrestricted access to powerful AI tools simply because the technology is available?
The European Commission is expected to formally present its proposed EU Kids Act this week. According to a European Commission document seen by Reuters, the plan would restrict children under 15 from accessing social media, video-sharing platforms, AI chatbots and online gaming services. The proposal would introduce different levels of access according to age rather than treating all children as one group.
Under the proposed framework, children aged 15 and above would be able to establish their own accounts. Those aged 13 and 14 could potentially use introductory accounts with parental approval, limited contacts and strict time controls. Children between 3 and 12 would be limited to parent-controlled, child-friendly services, while children under three would have no access to the covered services.
The significance of the EU proposal lies in its inclusion of AI chatbots and assistants, not just conventional social-media platforms. Services such as ChatGPT could fall within the scope of the proposed rules. Platforms would also be expected to implement age verification, parental controls, mechanisms for reporting harmful material and measures against addictive design and harmful recommendation feeds.
The EU’s approach reflects a growing concern that children’s digital risks are no longer limited to excessive social-media scrolling. AI systems can answer questions, generate convincing information, imitate conversations and increasingly function as personal assistants or companions. For younger users, that creates questions about dependence, misinformation, privacy, emotional influence and whether children can adequately judge the reliability of an AI-generated answer.
That debate is already being tested much more directly in New York City schools.
New York City Public Schools has introduced a one-year moratorium on student-facing generative AI for students from 2-K through Grade 8 during the 2026-27 school year. Companion chatbots are prohibited across all grades, while high-school students can use only limited, approved and vetted AI programmes. The policy affects nearly 600,000 students, roughly two-thirds of the city’s public-school population.
The policy is notable because New York has not declared that AI itself is bad or that students should never learn to use it. Instead, the city is drawing a distinction between learning about AI and allowing AI to perform the learning process for children.
Younger students are being given more space for traditional instruction, reading, writing, problem-solving and what educators describe as productive struggle. New York officials argue that children need opportunities to make mistakes, work through difficult problems and develop their own reasoning rather than immediately turning to an algorithm for an answer.
New York is also restricting screen exposure. The city’s guidance recommends no more than 30 minutes of daily one-to-one screen time for students in Grades 3-5 and 45 minutes for Grades 6-8, with exemptions for certain educational, assessment and assistive technologies.
That distinction could be particularly important for India, where AI tools are rapidly entering classrooms, homework routines and private tutoring. The central question for Indian schools should not simply be whether students are using ChatGPT or another chatbot. It should be whether the technology is actually improving learning — or quietly replacing the cognitive effort through which children develop writing, mathematics, reasoning and independent judgment.
India has a very different education system from New York, with enormous differences between urban private schools, government schools, coaching centres and rural classrooms. A blanket ban therefore may not be practical or desirable. But New York’s policy offers a useful warning against allowing AI adoption to run ahead of educational evidence.
Schools in India could consider adopting an age-based AI policy rather than allowing unrestricted use across all grades. Younger children could receive stronger restrictions, while older students could gradually gain access to AI tools alongside lessons on verification, privacy, plagiarism, hallucinations and responsible use.
There is another lesson in New York’s approach: schools should evaluate the technology before purchasing it. NYC Public Schools says its technology review will consider vendor accountability, safety and transparency, evidence of learning design, instructional impact, research and continuing user feedback. Tools that are not considered mission-critical to learning can be barred.
This could be especially relevant in India, where schools are increasingly being marketed AI-enabled educational products. The presence of an AI feature does not automatically make an educational application better. Schools need evidence that the tool improves comprehension, retention, writing or problem-solving rather than merely making lessons more entertaining or reducing teachers’ workloads.
The EU’s proposed rules also highlight another issue Indian policymakers will eventually have to confront: age verification and child privacy. If governments require technology companies to determine whether users are children, they must also consider how that information is collected, stored and protected.
At the same time, a complete prohibition has its own limitations. Children will continue encountering AI outside school through smartphones, family devices and the wider internet. New York itself acknowledges that its school policy cannot prevent students from using AI at home. Its education officials therefore hope the restrictions will encourage parents and children to discuss responsible AI use beyond the classroom.
For India, perhaps the strongest lesson is therefore not “ban AI in schools.” It is do not confuse access to AI with AI literacy.
A student who can ask a chatbot to write an essay may be using AI. A student who can identify a fabricated citation, challenge an incorrect answer, protect personal information, compare AI output with reliable sources and explain the reasoning independently is becoming AI-literate.
That distinction will become increasingly important as AI becomes embedded in education. The emerging policies in New York and Europe suggest that governments are moving toward a more cautious model: protect younger children, impose stronger safeguards, make technology companies accountable and introduce AI gradually rather than treating every new AI capability as automatically suitable for the classroom.
For Indian schools, the message is timely. The race should not be to put AI into every classroom. The priority should be deciding where AI genuinely helps children learn — and where it may prevent them from learning at all.
