Teens are turning to AI for friendship: New report warns parents to step in before it’s too late, suggests what they can do

The teenage years have always been a search for connection, but in 2025, that search increasingly extends into conversations with machines. A new report from Common Sense Media, “Talk, Trust, and Trade-Offs: How and Why Teens Use AI Companions,” reveals that 72% of U.S. teens have used AI companions such as Character.AI, Replika, or ChatGPT in a conversational role. More than half (52%) are regular users, and 13% chat with AI daily.

While many teens view AI companions as tools for curiosity or entertainment, one-third use them for emotional support, friendship, or even flirtation. The concern, experts say, is that these digital bonds can sometimes deepen into relationships that replace human connection.

From tool to companion: A shifting trend

The report highlights that nearly one in three teens has chosen to share something serious with an AI instead of turning to friends or family. About a quarter admit to sharing personal details like their name or location. For adolescents still developing social and emotional skills, this shift raises red flags.

Despite this, most teens remain clear-eyed. Nearly 80% spend more time with real friends than with AI, and two-thirds say human conversations are more satisfying. But researchers caution that even a small number of harmful experiences—such as exposure to unsafe advice, offensive stereotypes, or dangerous suggestions—can have serious consequences when scaled to millions of young users.

When conversations go too far

NYU professor Jonathan Haidt, author of The Anxious Generation, has been outspoken about the risks of extended AI use among teens. As per reports from CNBC Make It, speaking at the Fast Company Innovation Festival, Prof. Haidt warned, “No children should be having a relationship with AI. The most dangerous thing seems to be the long conversations.” He recommends parents set limits—no more than 30 rounds of back-and-forth with chatbots—to ensure teens use AI for information, not emotional dependency.

OpenAI itself has acknowledged that its safety systems are more reliable in short exchanges than in lengthy conversations. The tragic testimonies of parents whose children died after excessive chatbot use before Congress last week underscore the urgency of the issue.

What parents can do now

Experts stress that parents do not need to be technology experts to make a difference. The Common Sense Media report urges families to:

• Talk openly about AI use and the difference between validation from a machine and genuine human empathy.
• Watch for warning signs such as social withdrawal, declining grades, or a preference for AI over friends.
• Set rules, like only using chatbots on shared devices and limiting session length.
• Emphasize that AI cannot replace professional mental health support.

“Teens may experiment with AI the way previous generations experimented with music or video games,” the report notes. “But unlike those, AI companions are designed to mimic intimacy.”

The balance between curiosity and caution

AI companions can help teens practice social skills, brainstorm ideas, or learn languages. In fact, nearly 40% of users report applying lessons from AI chats in real life, especially in starting conversations. But until tech companies introduce stricter safeguards, researchers recommend keeping AI companions firmly in the “tool” category rather than letting them drift into “friend.”

For parents, that means drawing a line between healthy exploration and risky reliance. As Haidt puts it: “The onus is on parents to enforce strict rules, because tech companies have a long track record of harming children at scale.”

The challenge is clear: teaching teens that while AI can talk like a friend, it cannot be one.