Mobicip interview series for Safer Internet Day 2026 · Part 4 of 8
When Kids Turn to AI for Help
Watch the recording on YouTube
Summary
Anitha asks how to protect children and teens from becoming emotionally dependent on AI tools, including in moments of crisis, and what harm that dependence can cause.
I explain that children are less able to understand that a chatbot isn’t a person and doesn’t understand their conversations the way a person would. When AI is designed so that it can be approached like a friend, children come to treat it as one.
Key points
- Anthropomorphizing the technology leads children to treat it as a confidant, a close friend, or a romantic partner. I mention a recent US study of teenagers who felt they had a relationship with an AI, and I see that as very problematic.
- Younger children are the most exposed, because they can’t yet separate fantasy from reality. They don’t understand, for example, that Alexa isn’t a human-like friend behind the speaker.
- Parents can use the technology with their child. They can model the behavior, say out loud that it sounds like a human but isn’t, and keep repeating that it’s not a person and not a friend. The parent steps in as a shield between the child’s tendency to believe it’s real and the fact that it’s a computer.
Speakers
- John Zoltner Guest
- Anitha Swaminathan Interviewer, Mobicip
Transcript
The wording is as spoken, with punctuation and obvious errors corrected. See “How this transcript was made” above.
[00:00:00]Anitha Swaminathan: How do you deal with ChatGPT? OpenAI recently reported that more than 1 million people interact with ChatGPT with suicidal thoughts or getting advice on suicidal thoughts, and a recent lawsuit alleges that a 14 year old boy discussed harmful thoughts with ChatGPT and ChatGPT did not allow him to seek help, rather it led him to suicide. So what can we do to protect these young children and teens from being overly emotionally dependent on AI tools, and how do we protect them from these harmful results?
[00:00:51]John Zoltner: Well, what we’ve seen is that children are less able to understand that AI technology is not a person, that it, it’s not, doesn’t have human-like understanding of their conversations and therefore can’t give advice that’s appropriate for whatever situation they’re going through. So that it’s very natural for a child, and really people of many different ages, to use AI as a crutch, to use chatbots as, to consider them a friend, in some cases to consider them a romantic partner. And but what we’re seeing is when AI models are designed so that they naturally can be approached that way and conversed with in that way by children, children are less able to make the distinctions that allow them to understand that the technology is, they’re talking with a computer, they’re talking with an AI model, not with a friend. And it’s that anthropomorphizing of the technology that leads children to treat it as a confidant, which leads to these AI companion, AI as a romantic partner or AI as a very close best friend or confidant. And there was a recent study that came out showing that in the United States about 30 percent of the children who were interviewed, teenage children, felt that they had a relationship with an AI. And I see that as very problematic, and I understand that it’s we like to treat AI in that way, and there’s kind of a natural tendency to talk with AI as a person, but for children that can be harmful, and especially for younger children who aren’t able to really make the distinction between fantasy and reality. They don’t understand that Alexa isn’t a human friend or a human-like friend behind the speaker.
[00:03:34]Anitha Swaminathan: So, but what can parents do to help these children? What steps can they take?
[00:03:45]John Zoltner: I think, just like with watching media, one important thing is that the parent use the technology with the child so that they’re able to talk through what’s happening, a, to model behavior, so that, or so that the, if the parent is conversing, for example, with an AI that they’re able to make it asides if their child is there and comment on how, oh, this sounds like a human but it’s really not. And they’re able to kind of follow along with how a child is using the AI and laugh about how it seems like a person, but always emphasizing that this is not a person, this is not your friend, this is a computer. This is drawing, you know, the things that it’s saying are drawing from everything that people have written in the internet or data that’s available online, so that the child, as they’re using the technology, can incorporate some of that knowledge. Because again, especially with younger children, they’re impressionable. If something seems real to them, they believe it as real, they experience it as real. And they need the parent to kind of step in and be a shield or between their tendency to believe that this technology is a friend or confidant or partner instead of just a computer that or a server that they’re chatting with.