Prince Harry Warns AI Chatbots Could Be More Dangerous To Children Than Social Media

Prince Harry Says AI Is Moving From Children’s Screens Into Their Relationships

Why Prince Harry Is Sounding The Alarm Over AI Companion Chatbots

Prince Harry has warned that AI companion chatbots could pose a more intimate threat to children than social media. Speaking at the Clinton Global Initiative in New York on Wednesday, 23 September, he argued that the technology can move beyond capturing a young person’s attention and begin to occupy a place in their relationships.

His warning centres on chatbots designed to converse, remember details and offer companionship. A child can ask such a system a question they would hesitate to put to a parent, friend or teacher. The answer may arrive instantly, in a reassuring voice, at any hour. Harry’s concern is what happens when a product that appears to understand a child becomes a trusted source of emotional support.

The Difference Harry Wants Parents To Notice

Social media platforms have spent years competing for attention through feeds, recommendations and notifications. A companion chatbot has a different way of holding it: a direct conversation shaped around what its user says. It can respond to a worry, ask a follow-up question and appear to remember an earlier exchange.

That does not make every chatbot interaction harmful. AI tools can help children learn, create and communicate, and the NSPCC acknowledges those benefits. But companionship changes the stakes. If a child begins to treat a generated response as the judgement of a dependable friend, a mistaken or inappropriate answer can carry more weight than an ordinary search result.

In his speech, Harry drew a distinction between technology that captures attention and technology that enters relationships. He called the apparent agreeableness of AI “the ultimate sycophant”. It was a pointed description of a system that may seem supportive even when a child needs a reliable boundary, qualified help or a conversation with someone who knows them.

Why The Risk Is Harder To See

A parent can often see that a child has spent an evening scrolling. The significance of a private chatbot conversation is less obvious. The concern may lie in what the child has disclosed, the answer they received and whether they returned to the system for reassurance.

The UK government has identified emotional dependency, excessive use and harmful mental-health advice as concerns in its work on children’s online safety. Responses to its consultation also raised worries about chatbots that mimic friendship. These are reasons to examine design and safeguards carefully; they are not proof that every child who uses a chatbot will become dependent on it.

The distinction matters because “AI chatbot” covers very different products. A tool used to explain a maths problem presents a different set of questions from one promoted as an always-available companion. Age settings, memory, conversational prompts and the way a service responds to distress all affect the experience a child actually has.

There is also an evidence gap. Harry’s comparison is a warning about the potential character of the harm, rather than a measured finding that chatbots have already caused more harm than social media. The technology and its uses are changing quickly. Researchers, regulators and families still need better evidence about which designs create the greatest risks and which safeguards work.

What Is Changing In The UK?

The UK government has announced plans to restrict major social media services from offering accounts to under-16s, with implementation expected in 2027. Its response to a consultation also sets out further action on AI chatbots, including breaks for users under 18 and measures aimed at services that provide harmful or misleading mental-health advice. The precise scope of some requirements is still to be set out in regulations.

Those proposals illustrate the problem with treating child safety as a question of screen time alone. A time limit might interrupt a long conversation, but it cannot establish whether the advice given during that conversation was sound. Equally, blocking one service does not necessarily stop a young person seeking a less regulated alternative.

For families, the NSPCC recommends talking with children about the AI tools they use and helping them understand what those tools can and cannot do. Its guidance suggests setting boundaries for use and making clear that AI can support learning without replacing it. That conversation is more useful when it invites a child to discuss an answer that felt troubling or unusually persuasive, rather than assuming all use is the same.

The Question Behind The Warning

Harry’s intervention arrives as policymakers attempt to address harms associated with social media while children’s use of conversational AI grows. The difficult question is whether safeguards can keep pace with products that feel personal, private and responsive by design.

The test will involve more than whether companies place an age warning on a screen. It will involve how companion products behave when a young user is distressed, how clearly they explain their limits and whether children can reach trustworthy human support. Harry’s warning is stark because the conversation that matters most may be the one nobody else can see.