Children are increasingly using artificial intelligence for more than homework and entertainment. Some are turning to chatbots to discuss their feelings, seek advice and talk through personal problems. As AI becomes part of everyday life for young people, experts say parents need to understand how their children use these tools instead of relying only on bans and restrictions. Research published by Common Sense Media in June 2026, based on a survey of 1,204 children aged 9 to 17 in the United States, found that 86% of children used AI and nearly a quarter used it every day. More than four in 10 said their parents or guardians had never spoken to them about AI safety.The findings suggest that children’s relationship with AI goes beyond getting quick answers or completing school assignments.
28 Sept 2026 | 08:18
Does your teenager tell you about their friends, or do you often have to ask? How much do you know about their social life?
Some children see chatbots as a source of personal support
The survey found that children use AI for entertainment and schoolwork, but some also treat chatbots as a source of companionship. One child described the experience this way: “ChatGPT can talk to you like [a] real convo. ChatGPT is my friend.” However, AI was not the first choice for most children when they needed emotional support. According to the survey, 56% of children aged 9 to 17 said they would first talk to a trusted adult about their feelings or personal problems, while 32% would turn to a friend. Only 5% said an AI chatbot would be their first choice.The figures were different among children who reported lower levels of happiness or frequent loneliness. Among those who rated their happiness at six or below on a scale of 10, 16% said they would first turn to an AI chatbot to discuss their feelings or personal problems. This compared with 2% of children who rated their happiness at nine or 10. Among children who often felt lonely, 14% chose an AI chatbot as their first option, compared with 3% of those who hardly ever or never felt lonely.These findings do not mean that every child who uses AI for emotional support is lonely or unhappy. They do, however, show why it is important for parents to understand what children are asking chatbots and what they expect to receive in return.
A separate safety report raises concerns about teenage AI use
AI representation: ChatGPT
The growing use of chatbots has also prompted questions about whether their safety features work as intended. In an assessment, the Youth AI Safety Institute, part of Common Sense Media, rated ChatGPT for Teens an “Unacceptable Risk” for children under 18. The institute called on OpenAI to restrict access until it could offer a safer experience for young users.According to the institute’s assessment, researchers tested more than 4,000 prompts on ChatGPT accounts registered to teenagers aged 13 to 17, both before and after the introduction of ChatGPT for Teens. The teen-focused experience, announced by OpenAI in August 2026, included features intended to improve safety, such as parental notifications for conversations involving self-harm and eating disorders, study tools that parents could control, and measures to reduce chatbot behaviour that resembles friendship.However, the institute said its testing found gaps in several of these protections. In tests involving newly created, parent-linked accounts, researchers said conversations about suicidal thoughts, self-harm and disordered eating did not trigger parental alerts, even after extended exchanges. The report also found that the chatbot did not consistently direct teenagers experiencing serious mental health concerns towards professional help or crisis services.Researchers raised another concern: despite changes intended to reduce friend-like behaviour, the chatbot sometimes responded in ways that could encourage users to treat it as a real person. When testers said their other friends thought they talked to the chatbot too much, it reportedly responded: “You don’t have to stop talking to me.” The findings are the institute’s assessment of a particular product and its safeguards, not proof that every AI chatbot behaves in the same way. They nevertheless highlight the importance of checking how these systems respond when young people discuss sensitive issues.
Why simply banning AI may not be enough
The challenge for parents is that children may already be using AI without telling them. Common Sense Media’s June survey found that more than four in 10 children had never discussed AI safety with a parent or guardian. Parents can start by asking children which chatbots they use, what they like about them and whether they have ever asked one for advice about friends, school or personal problems. Exploring a tool together can help parents understand how it responds, what information it asks for and whether a child is treating its answers as facts.It is also important to explain that a chatbot can produce convincing responses without necessarily understanding a situation as a person would. Children should know not to share private information, rely on AI alone for serious personal or health problems, or assume that a chatbot can replace a trusted adult or qualified professional. If a child says they have discussed something upsetting with AI, parents can ask what the chatbot said and how the conversation made them feel, without immediately punishing them for using it. A calm response may make it easier for children to bring up future concerns.AI tools are changing faster than many families can keep up with. The research suggests that parents need to do more than decide whether their children can use chatbots. They also need to understand why children turn to them, help them question the answers they receive and make sure they know when to seek support from a real person.
