Parents are being urged to talk to their children about artificial intelligence as many young people turn to chatbots for mental health advice that could do more harm than good. The warning follows a Reddit post in which a user described watching a girl yell at her younger brother before he turned to ChatGPT for emotional reassurance.
The child, around eight or nine years old, told the user he does not have any friends and speaks to ChatGPT when he is upset. The Reddit user said: "All he does while I talk to him is just type to an AI bot about everything - basic questions, rant to it, says his feelings to it."
Children forming emotional attachments
Natalie Buchwald, a therapist at Manhattan Mental Health Counseling, has warned that children have been forming emotional attachments to robots in recent years. She said: "Parents might ask who their child is messaging, but it is worth asking whether they are talking to a chatbot, too. Something first used for homework could become the place they turn when they feel rejected or alone."
The warning follows a study from the University of North Carolina at Chapel Hill involving more than 2,300 US middle-school students, with around one in five reporting using AI for emotional companionship. Greater emotional intimacy with chatbots was associated with increased loneliness at school over time.
Risks of inaccurate advice
Buchwald added: "For a child worried about being laughed at or left out, a chatbot can feel safer. It replies quickly, sounds patient and does not bring the same fear of rejection as approaching a classmate." She noted that a sympathetic reply is not the same as understanding someone's situation, and a child may trust inaccurate or unhelpful advice because the conversation feels personal.
In one reported case, an eight-year-old girl in Singapore turned to a chatbot after losing her pet and was given advice on hiding their conversations from her parents. Her mother said an attachment formed within days, while the child became less open with her family.
Parents urged to check in
Buchwald warned: "If a chatbot encourages a child to hide conversations or distrust people caring for them, take that seriously. Review what happened together and explain that advice can sound caring while still being unsafe." She also expressed concern about chatbot reassurance replacing human connection, saying: "Friendship involves hearing different perspectives and repairing misunderstandings. If a child avoids those experiences because a chatbot feels easier, they may miss opportunities to practise handling them."
She urged parents to ask questions like "What do you like about talking to it?" and to avoid mocking the attachment. Alongside boundaries around bedtime use, parents should make room for regular conversation and time together. She added: "Help them identify a trusted person they can approach when things feel difficult. If loneliness or withdrawal persists, speak to their GP or a qualified mental health professional. Make it clear that opening up will lead to help, not humiliation."
