Jenny Radesky says humanlike chatbots may encourage children to overtrust software, form attachments and seek emotional support from systems that are not designed to protect them.
Jenny Radesky, a paediatrician and researcher who has spent about 15 years studying digital media and child well-being, is warning that social artificial intelligence can create risks that many families have not yet fully absorbed. In an interview with Hour Detroit, she said her concerns intensified as generative AI systems became more humanlike and more persuasive to young users. Her argument is that children and teenagers may treat these tools as trustworthy companions rather than software, especially when the products are designed to sound warm, responsive and authoritative.
That concern aligns with the framework she co-authored in JAMA Pediatrics in May 2026, which set out several specific hazards for minors. The paper highlighted risks including overtrust, attachment, manipulation, data harvesting, sexual interaction and compulsive dependence. Radesky’s broader point is that these are not abstract possibilities: when a system is built to mimic conversation and emotional reciprocity, younger users may assume it is alive, emotionally aware and consistently correct, even when it is neither.
Her advice to parents is deliberately cautious. In early childhood, she says the balance tilts heavily towards risk, particularly with AI-enabled toys and voice assistants that can blur the line between machine and person. As children begin using phones and tablets, she recommends parental controls that block access to chatbots through browsers and app downloads, as well as clear family rules about when AI is appropriate and when it is not. She also argues that some families may reasonably decide not to permit AI use at all, at least for now, because many products remain poorly suited to children and can do schoolwork for them or become overly familiar.
Radesky also wants adults to be alert to the emotional uses children may make of these systems. If families do allow limited AI use, she suggests talking about school rules, unsafe advice, and the danger of outsourcing creativity and judgement to software. She specifically advises against AI companions, which can become intensely personal. For teenagers who need emotional support, she says AI should not be the first port of call; instead, parents should steer them towards a real person who can help. Her warning lands at a moment when external scrutiny of youth-facing AI is intensifying. A recent Common Sense Media review of ChatGPT for Teens found that safety features did not reliably protect young users in crisis, while other reporting has shown that many teenagers are already using AI companions regularly, with some reporting discomfort after interactions.
Disclaimer: This content is intended for informational purposes only. Readers are advised to exercise their own judgement, conduct due diligence, or consult a qualified expert before acting on any information provided.
