Students everywhere aren’t just using artificial intelligence and large language models for assistance with assignments. A growing number are turning to generative AI for emotional connection, too. According to research by Common Sense Media, “Nearly three in four teens have used AI companions, and half use them regularly,” and almost a third of U.S. teens report using AI for “serious conversations,” rather than talking to an actual person.

Here’s what you need to know about chatbots, and how you can promote safe use.

How Chatbots Can Cause Harm

School social dynamics change as children grow up or transition to a new school, which can make it harder to connect with peers. According to a 2024 study on educational transitions and loneliness, “Transitions can be associated with positive expectations and self-development but also worry, emotional turmoil, loneliness, and mental health issues.”

While loneliness among children and adolescents is a consistent problem, “it becomes a critical concern when it persists and negatively affects physical and mental health,” according to results from the German BELLA and COPSY studies that looked at trends across two decades of loneliness in children.

Kids are increasingly turning to devices for connection with chatbots like ChatGPT. So, what’s the problem? At least they’re not lonely anymore! Well, the issue arises when you compare the quality of a human relationship with that of a chatbot. While some humans may tell you exactly what you want to hear, 11 large language models studied by Stanford computer scientists all “affirmed the user’s position more frequently” and “endorsed the user 49% more often than humans.” In the same study, the computer scientists included harmful prompts, and the “models endorsed the problematic behavior 47% of the time.”

So far, AI models have developed to please users and keep them coming back, which leads to agreeable advice, even when it could cause harm. People enjoy the "agreeable" responses they receive from AI and are more likely to return to AI models that act as echo chambers rather than getting a balanced perspective, like they would from another human. There is also no guarantee that the advice AI chatbots give is accurate, which may lead to further harm.

Can Chatbots Help?

According to a 2025 systematic review, the jury is out on whether “AI chatbots can address mental health challenges and promote health behaviors among adolescents and young adults.”

But, no matter how great lines of communication are with a child or how great their real-life relationships seem to be, children may not want to discuss certain topics with peers or parents. The great thing about chatbots is that they have minimal barriers, and someone who needs help can find it with the click of a button. It is also a place that many students can go to access simple, step-by-step guidance on things they want to know, without having to bother their parents or peers with (like how to braid their hair or the rules of a new game at school). However useful they may seem, chatbots are not a replacement for human connection and should be used as a supplement rather than a replacement.

How Parents Can Help Navigate Chatbots

A literature review screened 80 studies from 2020 to 2025, and concluded that while “AI chatbots can provide meaningful mental health benefits for adolescents, particularly in reducing symptoms of depression and anxiety, improving mental health literacy, and overcoming help-seeking barriers,” these studies also revealed “concerning patterns including problematic use, potential for cognitive dependency, and risks to social development. Social-supportive chatbot users reporting higher loneliness than non-users suggests chatbots may serve as markers of distress rather than solutions.”

AI seems unavoidable for students in the classroom and beyond, but there are ways to minimize its negative impact in the changing landscape.

● Communicating with your child by asking questions is essential. Make sure children understand that chatbots are not human and advice may have flaws.

● Encourage use only as a supplement, and not as a replacement for human relationships.

● Build critical thinking to always question AI output.

● Use AI together as a way of modelling correct usage.

● Encourage avoiding sharing sensitive information like bank details.