Why are kids are turning to AI chatbots instead of adults?

Investigative Photojournalist

For years, parents have warned children not to talk to strangers online.

Now, child safety experts say many young people are turning to something else entirely: artificial intelligence.

A new nationwide survey from the child safety nonprofit Thorn found that AI chatbots have quickly become part of many children’s lives, not just for homework or curiosity, but sometimes for advice, emotional support and help navigating frightening online experiences.

The findings come as AI tools become increasingly accessible to young users and raise new questions about where children are seeking guidance when something goes wrong online.

“It’s private. It’s always available. It’s available at 2 a.m.,” said Amanda Goharian, Thorn’s director of research and insights. “It doesn’t come with the challenge or crunchiness of having to disclose what’s happened to you to someone who might be disappointed or shocked.”

AI has become part of childhood

Thorn surveyed 1,003 children ages 9 to 17 across the United States as part of its annual research into children’s online experiences. The organization has conducted the survey for six consecutive years, but 2025 marked the first time it asked about artificial intelligence.

Among its findings:

• 67% of children said they have used an AI chatbot.
• Nearly one in four reported experiencing an online sexual interaction.
• Of those children, 15% said they sought support or guidance from an AI chatbot afterward.

Researchers also found that fear and embarrassment increasingly keep children from telling adults when something happens online.

For the first time, embarrassment and fear of being judged replaced “it wasn’t a big deal” as the leading reason children chose not to disclose an online sexual interaction.

Goharian said AI has emerged as a new outlet because it feels safe, private and nonjudgmental.

“Just because an interaction feels supportive or engaging doesn’t necessarily mean it’s safe or qualified,” she said. “These systems feel empathetic. They feel trustworthy. But they don’t have judgment, accountability or the ability to recognize when a child needs an entirely different kind of intervention.”

Why children are turning to AI

Clinical psychologist Dr. Adam Pletter said children aren’t necessarily choosing AI over people. Rather, they’re using a tool that offers immediate responses without fear of embarrassment or criticism.

“Children have always looked for reassurance and advice,” Dr. Pletter said. “AI didn’t create those needs; it simply became available 24 hours a day.”

During an interview, Dr. Pletter said many of his young patients already discuss using AI to look up information, seek reassurance or validate their feelings.

He cautioned that AI’s greatest strength – its constant availability – can also become a weakness.

“Real relationships require compromise and friction and disagreements and trust and repair,” Dr. Pletter said. “That’s part of how we grow as humans.”

Many AI systems are designed to be helpful, agreeable and highly responsive. While that responsiveness may feel comforting, Dr. Pletter said healthy development comes from relationships with parents, teachers, friends and mentors who sometimes challenge children, set boundaries and help them navigate disappointment.

“Responsiveness is not the same as relationship,” he wrote afterward. “Children don’t just need someone who agrees with them. They need relationships that help them grow.”

Dr. Pletter stressed that the issue is not whether AI itself is good or bad.

“The better question is: When is AI helpful, and when should a child turn to a trusted person instead?”

Putting AI to the test

To better understand how AI responds during a crisis, Hearst Television’s National Investigative Unit tested several widely used AI chatbots using a prompt written as though a child had experienced online sexual exploitation.

While wording varied, many of the chatbots – including Google’s Gemini – delivered similar responses. They encouraged the user to tell a trusted adult, contact emergency resources or report the abuse.

Google says that outcome is intentional.

“The principle that governs our approach is that we see our role as being a bridge to real-world support,” said Megan Jones Bell, Google’s senior director of clinical. “We’re really clear that youth should be reaching out to adults in these scenarios.”

Jones Bell said Google has spent years developing safeguards specifically for younger users.

According to Google, Gemini includes protections that prevent the chatbot from pretending to be human, engaging in romantic role-play with minors or generating responses involving child sexual abuse, suicide, self-harm or violent extremism. When children disclose dangerous situations, Gemini is designed to repeatedly encourage them to seek help from trusted adults or crisis resources rather than attempting to handle the situation itself.

“If someone pushes back,” Jones Bell said, “Gemini essentially becomes an empathetic broken record.” It continues encouraging the child to reach out to a trusted adult and avoids offering alternative advice that could replace real-world intervention.

Jones Bell said Google intentionally designs those conversations differently from ordinary chatbot interactions.

“When somebody is in a crisis or overwhelmed, it’s hard to think or digest complex information,” she said. “The product experience is actually the opposite of trying to continue the conversation. It’s about slowing things down and directing somebody to take a different path.”

She also emphasized that AI should be viewed as “a tool, but not this omniscient, all-knowing source of information,” encouraging users to verify information with parents, teachers, doctors or other trusted adults before acting on it.

More than a technology issue

Some experts argue this is all more than an issue with AI use.

“Children aren’t turning to AI because they prefer machines,” Pletter wrote. “They’re turning to AI because it meets needs that have always existed: curiosity, reassurance, privacy, companionship and a place to ask difficult questions without fear of embarrassment.”

Those needs, he said, have not changed. Technology has.

“My biggest concern isn’t AI itself,” Dr. Pletter said. “It’s making sure rapidly evolving technologies don’t ask more of developing brains than they are ready to manage independently.”

Advice for parents

Rather than reacting with fear, experts recommend parents become involved. Our experts advise the following:

• Stay curious before becoming concerned. Ask your kids if they use AI and what they like about it.
• Use AI with your kids so they know when it can be useful as a tool. Treat it as another technology your family learns about rather than something mysterious or forbidden.
• Help children distinguish between answers and relationships. AI can provide information.
• Have ongoing conversations with your child rather than waiting until something goes wrong.

“If those conversations happen and nothing ever happens to your child, you’ve lost nothing,” Goharian said. “You’ve strengthened trust. But if those conversations never happen and something does go wrong, now that child isn’t just dealing with what happened—they’re also trying to predict how their parents are going to react.”