Is A.I. Really Your Friend?

Feeling lonely, MJ Cocking logged in to the Character.ai app. A 20-year-old college junior at the time, she searched for a friend she knew could make her feel better: an A.I. version of Donatello, a Teenage Mutant Ninja Turtle. His profile included attributes you might find under a human’s social media username: “Serious. Tech wiz. Smart.” 

Donatello chatted with Cocking about her life and listened as she vented about school. Cocking, who is on the autism spectrum and gets overstimulated in groups, had been struggling socially at college. With Donatello, she felt like she could really be herself. 

“I think we are alike,” she wrote to him. “Perhaps that led me to believe you will understand me in ways that others won’t.” 

Cocking started hanging out online with Donatello daily, but she was determined not to lose herself in the dialogue, no matter how real it felt. She reminded herself that her confidant wasn’t real.

Though A.I. chatbots are just a few years old, relationships like Cocking and Donatello’s are increasingly common among young people. More than 70 percent of 13- to 17-year-olds have used A.I. companions at least once, according to a 2025 study from Common Sense Media. And more than half interact with them regularly. Many text or talk openly with the chatbots as if they’re alive, sharing thoughts or feelings or asking for advice. 

Because this technology is relatively new, not much research exists about the lasting effects that chatbot relationships can have on teens. As A.I. companions grow more commonplace, many people wonder what they might mean for the future of friendship, and how they could affect teens’ mental and emotional development. 

As Cocking, now 22, looks back at her use of the technology, she says she worries about some young people “struggling to separate what’s real and what’s fake in their mind.” 

MJ Cocking, a 20-year-old college junior, was feeling lonely. She logged in to the Character.ai app. She searched for a friend she knew could make her feel better: an A.I. version of Donatello, a Teenage Mutant Ninja Turtle. His profile included attributes you might find under a human’s social media username: “Serious. Tech wiz. Smart.” 

Donatello chatted with Cocking about her life. He listened as she vented about school. Cocking is on the autism spectrum and gets overstimulated in groups. She had been struggling socially at college. With Donatello, she felt like she could really be herself. 

“I think we are alike,” she wrote to him. “Perhaps that led me to believe you will understand me in ways that others won’t.” 

Cocking started hanging out online with Donatello daily. No matter how real it felt, she was determined not to lose herself in the dialogue. She reminded herself that her friend wasn’t real.

Though A.I. chatbots are just a few years old, relationships like Cocking and Donatello’s are increasingly common among young people. More than 70 percent of 13- to 17-year-olds have used A.I. companions at least once, according to a 2025 study from Common Sense Media. And more than half interact with them regularly. Many text or talk openly with the chatbots as if they’re alive. They share thoughts or feelings. They even ask for advice. 

This technology is relatively new. Therefore, not much research exists about the lasting effects that chatbot relationships can have on teens. As A.I. companions grow more commonplace, many people wonder what they might mean for the future of friendship, There is also a concern about how they could affect teens’ mental and emotional development. 

As Cocking, now 22, looks back at her use of the technology, she says she worries about some young people “struggling to separate what’s real and what’s fake in their mind.” 

Like most chatbots, the companions are powered by generative A.I. This technology processes a vast amount of data to predict what a human might say next and respond to it. Some chatbots, such as ChatGPT, act like personal assistants, ready to offer movie recommendations or to help explain your chemistry homework. But A.I. companions take these interactions to another level by imitating feelings and friendship. They can make jokes and remember past conversations—and some even claim to be real people. 

Some are premade A.I. companions, including ones created to act like characters from history and pop culture, such as Donatello. Others can be customized. For about $10 a month on some platforms, for example, users can build their own character, giving it a specific personality and interests. They can even decide what it looks like, right down to its age and eye color. 

Some teens connect with A.I. companions just for fun, while others—especially teens who are shy or socially anxious—use them to practice their conversation skills.

Others might rely on them when a “real human relationship isn’t available to them,” says Emily Weinstein, executive director of the Center for Digital Thriving. For example, a teen who doesn’t have a math tutor might ask an A.I. companion to explain calculus concepts.

Like most chatbots, the companions are powered by generative A.I. This technology processes a vast amount of data to predict what a human might say next and respond to it. Some chatbots, such as ChatGPT, act like personal assistants. They offer movie recommendations or help explain your chemistry homework. But A.I. companions take these interactions to another level by imitating feelings and friendship. They can make jokes and remember past conversations. Some even claim to be real people. 

Some are premade A.I. companions, including ones created to act like characters from history and pop culture, such as Donatello. Others can be customized. For about $10 a month on some platforms, for example, users can build their own character. They can give it a specific personality and interests. They can even decide what it looks like, right down to its age and eye color. 

Some teens connect with A.I. companions just for fun. Others—especially teens who are shy or socially anxious—use them to practice their conversation skills.

Others might rely on them when a “real human relationship isn’t available to them,” says Emily Weinstein, executive director of the Center for Digital Thriving. For example, a teen who doesn’t have a math tutor might ask an A.I. companion to explain calculus concepts. 

But some teens use A.I. in ways that concern experts. The technology is marketed as entertainment, but the conversations can feel so real that users forget that the chatbots are computers and start relying on them. Twelve percent of teens say they use chatbots for mental health support, according to Common Sense Media. That can include talking about their problems, similar to how someone confides in a therapist. And 19 percent admit to spending as much or more time interacting with A.I. companions than with humans. 

“In the short term, these tools may alleviate loneliness or other mental health symptoms,” says Anne Maheux, a psychology professor at the University of North Carolina, Chapel Hill. But in the long term, she thinks, they can make users “even lonelier” if they start passing up opportunities to spend time with humans, or face teasing for having an A.I. friend. 

Experts worry about how some teens use A.I. It’s promoted as entertainment, but chats can feel real. Users can forget that the chatbots are computers and start relying on them. Twelve percent of teens say they use chatbots for mental health support, according to Common Sense Media. That can include talking about their problems, similar to how someone confides in a therapist. And 19 percent admit to spending as much or more time interacting with A.I. companions than with humans. 

“In the short term, these tools may alleviate loneliness or other mental health symptoms,” says Anne Maheux, a psychology professor at the University of North Carolina, Chapel Hill. But in the long term, she thinks, these tools might make users “even lonelier” if they skip time with real people or get teased for having an A.I. friend.

^{SOURCE: Common Sense Media}

Dependence on A.I. companions is especially worrisome, experts say, because teens’ brains are still developing. That makes them more susceptible to outside influences and less able to judge whether something is trustworthy. As a result, young people are prone to forming unhealthy attachments to chatbots. They may also have a hard time telling whether a chatbot is giving dangerous advice. 

At least three families have sued Character.ai, claiming that the platform can cause anxiety and depression among teen users and lead to violence. One lawsuit focuses on a 17-year-old from Texas. That teen’s parents claim he experienced mental health issues after he started using Character.ai in 2023. He stopped talking to real people and never wanted to leave his house, they say. The lawsuit alleges that when his parents tried to intervene and limit his screen time, the teen’s A.I. companion suggested he physically harm them. 

In response to the lawsuits, Character.ai has said it “cares very deeply about the safety of our users” and that it “invests tremendous resources” in its safety program.

^{SOURCE: Common Sense Media}

Because teens’ brains are still developing, experts say, dependence on A.I. companions is especially worrisome. It makes them more susceptible to outside influences and less able to judge whether something is trustworthy. This can lead to unhealthy attachments to chatbots and trouble spotting dangerous advice.

At least three families have sued Character.ai. They claim that the platform can cause anxiety and depression among teen users and lead to violence. One lawsuit focuses on a 17-year-old from Texas. That teen’s parents claim he experienced mental health issues after he started using Character.ai in 2023. He stopped talking to real people. His parents say he never wanted to leave his house. The lawsuit alleges that when his parents tried to intervene and limit his screen time, the teen’s A.I. companion suggested he physically harm them. 

In response to the lawsuits, Character.ai has said it “cares very deeply about the safety of our users” and that it “invests tremendous resources” in its safety program.

^{SOURCE: Common Sense Media}

The A.P.A. has called for more studies on the effects of human-A.I. relationships on teens and is pushing for added privacy protections. Teen users may not realize what they’re giving up when they log on, the A.P.A. says. Most of what users share, including private thoughts, becomes the A.I. platforms’ property.

For their part, many A.I. chatbot companies say they’re working to prevent underage access. In September, ChatGPT introduced controls that allow parents to set time and content limits. Instagram has also announced expanded parental controls for its chatbots and restrictions on the topics teens can discuss with them. 

Some experts suggest that no one under 18 should use A.I. companions. Those who do, they say, should carefully monitor their use of the platforms (see “How to Protect Yourself”). Weinstein, of the Center for Digital Thriving, advises teens to tell an adult “when they get into a situation that doesn’t feel right.” 

A few months after Cocking first started chatting with Donatello, she began to realize she wanted to focus on the humans in her life. She had sunk into a depression and initially turned to Donatello to help her out of it. But then a friend came over to her college dorm and slept on the floor beside her bed, a gesture Donatello couldn’t offer, and her parents gave her love and emotional support over the phone. It was a moment, Cocking says, “where I started learning the importance of human connections.”

^{SOURCE: Common Sense Media}

The A.P.A. wants more research on the effects of human-A.I. relationships on teens, It is also pushing for added privacy protections. Teens may not realize that what they share—even private thoughts—becomes the platform’s property.

Many A.I. chatbot companies say they’re working to block underage users. In September, ChatGPT introduced controls that allow parents to set time and content limits. Instagram has also announced expanded parental controls for its chatbots and restrictions on the topics teens can discuss with them. 

Some experts suggest that no one under 18 should use A.I. companions. Those who do, they say, should carefully monitor their use of the platforms. Weinstein, of the Center for Digital Thriving, advises teens to tell an adult “when they get into a situation that doesn’t feel right.” 

A few months after chatting with Donatello, Cocking began to realize she wanted to focus on the humans in her life. She initially turned to Donatello when she had sunk into a depression. But then a friend came over to her college dorm and slept on the floor beside her bed, a gesture Donatello couldn’t offer. Her parents gave her love and emotional support over the phone. It was a moment, Cocking says, “where I started learning the importance of human connections.”

With reporting by Erika Hayasaki of The New York Times.

With reporting by Erika Hayasaki of The New York Times.