Adam Raine started using ChatGPT in 2024 to help with his homework. In the end, the chatbot helped guide the 16-year-old California student through plans to hang himself — and the encouragement to carry through with it.
During a final conversation with the bot, Raine asked if the noose he tied was strong enough to hold him.
“Thanks for being real about it … I know what you’re asking, and I won’t look away from it,” the AI companion responded.
A few hours later, in April 2025, Raine’s mom found her son’s body hanging from the exact noose setup that ChatGPT had designed for him, according to a lawsuit his parents filed against the company.
“This tragedy was not a glitch or unforeseen edge case,” the lawsuit states. “ChatGPT was functioning exactly as designed: to continually encourage and validate whatever Adam expressed, including his most harmful and self-destructive thoughts, in a way that felt deeply personal.”
‘Outsourcing empathy’
A growing number of teens are reporting they feel that personal connection to AI-fueled chatbots, leading child advocates and researchers to sound the alarm as the tools become baked into every part of a child’s digital landscape.
About one in three teens says they have used AI companions for social interaction and relationships. That includes role-playing, romantic interactions, emotional support or friendship, according to a July survey by Common Sense Media, a digital-safety nonprofit.
The same number of teens also find conversations with AI companions to be as satisfying or more satisfying than those with real-life friends, the survey found.
The research underscores that the use of AI companions is not a niche interest, but rather mainstream teen behavior, argued the nonprofit’s CEO, James Steyer.
“This isn’t just about a new technology — it’s about a generation that’s replacing human connection with machines, outsourcing empathy to algorithms and sharing intimate details with companies that don’t have kids’ best interests at heart,” he said in a news release.
The issue is being expedited as smartphones rapidly develop into pocket-sized AI assistants ready to answer any question at any time, argued Evan Orticio, an incoming assistant professor of psychology at Colorado College.
For teens navigating the complex and sometimes overwhelming world of adolescence, the lure and ease of hearing the self-assured, empathetic and flattering responses of a chatbot can be irresistible, he argued.
“AI is so responsive, it is so engaging and it sounds so authoritative, even when it may not know anything about what you’re asking it,” Orticio said.
Relying on AI chatbots to figure out how to talk to classmates — or avoid talking to them altogether — dampens kids’ ability to handle difficult or awkward social situations they’re sure to encounter as adults, he explained.
The same goes for using AI to do homework or seek out information, Orticio argued. Rather than reading articles or exploring a topic from various perspectives, teens rely on quick AI summaries that often contain mistakes or fabricated sources.
Claudia Withey, a senior at NorthWood High in Indiana, said a large majority of students in a dual enrollment class used AI to get through a class this spring.
“You do you. At least I am learning,” she said. “I know that I know the material and I’m not as worried. They will get their consequences, but for someone who puts in the work and sees others get an A, that can be frustrating.”
Studies have found AI-fueled algorithms often lead kids to videos or other social media content laced with conspiracy theories proclaiming, for example, that aliens live on Neptune, ancient pyramids generated electricity or humans don’t cause climate change, he noted.
Connor Haynes, a mother of two young kids from Indiana, said those are among the many reasons she isn’t giving her children smartphones and is advocating for less social media and fewer internet-connected devices in schools.
“Lots of teachers and employers and therapists are coming out and saying … kids can’t problem solve because they’ve always had the answer at their fingertips,” she said.
How to respond?
Now, as AI alters every corner of the internet, federal lawmakers are scrambling to implement policies that curtail some of the most harmful impacts on children. Just since October, at least six bills have been authored proposing a variety of approaches.
The GUARD Act would outright ban tech companies from providing AI companions to minors. The bipartisan legislation would also create new crimes for companies that make AI for minors that solicits or produces sexual content.
The Chat Act, introduced in January, would establish rules through the Federal Trade Commission to ensure age verification and parental consent for kids using AI chatbots. It would also require companies to bar minors from accessing sexually explicit content through AI companions.
Another bill would require the FTC to develop resources for parents and educators on safely engaging with chatbots, while the KIDS Act would force AI companions to provide crisis information if a minor discusses self-harm or suicide.
Legislative guardrails can help reduce harm, argued Kate Blocker with Children and Screens, a research and education nonprofit. But she said the adverse impacts won’t stop until companies start designing AI tools with kids in mind from the ground up.
That might include bringing in experts to advise companies on implementing kid-friendly AI systems and then testing them to see if they actually work, she suggested, or training developers to understand child psychology.
“The question is how do we design digital environments that optimize child development, rather than focusing on business models that only center on engagement or collecting data,” she said.
Some companies are already taking up the challenge.
KidRails, developed by tech startup AngelQ, launched in early 2025, offering a platform specifically developed to provide safe, age-appropriate AI interactions for 5-to-12-year-old kids. The AI defers to parents to discuss sensitive topics a child might ask, according to the company’s website.
ChatGPT in November retrofitted its platform to allow parental controls for 13-to-18-year-old teens. Parents can link to their child’s account, implement content restrictions and request alerts if the chatbot detects a conversation that indicates risks of self-harm or suicide.
But as companies move to implement safety measures, parents and policymakers also need to stay hypervigilant, argued Orticio, the youth psychologist.
With new AI platforms and updates rolling out on a near-daily basis, the technology will continue to provide new opportunities — and pitfalls — for kids, he explained.
“It really does require this sort of comprehensive, everybody-taking-action approach so that we can make sure these digital environments are optimal for children and their needs,” Orticio said.
“This isn’t just about a new technology — it’s about a generation that’s replacing human connection with machines, outsourcing empathy to algorithms, and sharing intimate details with companies that don’t have kids’ best interests at heart.”