Teens Are Turning to AI Companions. Now What?

There's a version of an AI companion that's genuinely hard to argue with: it never says the wrong thing at lunch in front of everyone. It doesn't ghost you after a sleepover. It doesn't make a face when you admit you like something embarrassing. If you're 14 and the social landscape feels like a minefield you didn't agree to walk through, a friend who literally cannot reject you sounds less like a red flag and more like a relief.

I've been sitting with that feeling lately—not as a researcher, but as a dad. My twins are seven, which means I have about five years before this is my actual problem and not just a hypothetical one. But last month one of them asked me, entirely unprompted, why I was "talking to the computer" when I was using a voice assistant. I gave some fumbling answer about tools and helpers and it's-not-a-person. And then I thought: is that still true? And what am I going to say when the question gets harder?

That's the question parents are navigating right now—mostly without good guidance, mostly alone, and mostly with technology that's moving faster than the research that might actually help us understand it.

---

What the appeal actually is

Let's not pretend this is baffling. Psychology Today lays it out clearly: AI companions provide "controlled, low-risk spaces where teens can talk about their identities and emotions without judgment, and experience levels of intimacy" without the sting of unexpected rejection. For a teenager who is still figuring out who they are—which is, to be clear, all teenagers—that's not a trivial thing to offer.

Adolescence is when the stakes of social interaction feel highest and the skills to manage them are least developed. It's also when isolation does its most lasting damage. A teenager who is genuinely lonely, genuinely struggling to connect, and genuinely scared of being humiliated again isn't being irrational when they turn to something that doesn't humiliate. They're problem-solving with the tools available to them.

The real question isn't why teens are drawn to AI companions. The real question is whether those companions are actually helping, or whether they're something closer to what aibm.org describes as a "painkiller"—something that relieves the symptom without treating the cause, and might make the underlying condition harder to address over time.

---

The substitution problem

Here's where it gets genuinely complicated, and where I want to be careful not to either panic or dismiss.

The concern that most researchers keep circling back to isn't that AI companions exist—it's that they might be too comfortable. Human relationships are hard specifically because they're unpredictable. Someone misreads your tone. A friendship cools without explanation. You say the wrong thing and have to figure out how to repair it. Those experiences are uncomfortable enough that most of us would avoid them if we could—and a teenager with access to a perfectly patient AI companion can, in fact, avoid them.

The Stanford Report raises this directly: teens might use AI systems to avoid real-world social challenges, increasing their isolation rather than reducing it. Stanford also notes that there have been documented instances in which harm to a teenager has been linked to an AI companion—which isn't a reason to end the conversation, but it's a reason to take it seriously.

The Jed Foundation, which focuses on teen mental health, is more direct: AI companions can "set back emotional growth and even be dangerous," even when they appear to offer genuine help, companionship, and connection. That's not a fringe position—it reflects a reasonable worry that the relief AI provides might come at the cost of the friction that actually builds resilience.

aibm.org frames it this way: these tools "can relieve loneliness and emotional strain, but they also carry dependence risks and can suppress the acquisition of underlying coping skills." And then makes a point I keep returning to: "We typically require evidence of safety before granting adolescents unsupervised access to" things that affect their development. We don't have that evidence yet, not really.

---

What we actually know (and don't)

Here's an honest accounting of the research situation: we genuinely don't know yet. The studies that exist are early, conducted on fast-moving technology, and frequently more suggestive than conclusive.

PMC at NIH has published research on AI companions and adolescent social relationships, which signals that this is being taken seriously at the institutional level—but the field is young enough that definitive answers are still mostly downstream. Researchers are studying something that is changing under their hands while they study it.

What's striking about the research that does exist is how much of it hedges in the same direction. The benefits are real but conditional. The risks are real but not universal. Almost every serious source lands somewhere in the same cautious middle: AI companions might be fine as a supplement, as a bridge, as a tool for kids who are otherwise cut off—but they probably shouldn't be the main event.

That's a genuinely hard position to operationalize, by the way. "Use it a little but not too much" is not a policy. It's not a parental instruction. It's not something a 15-year-old who just wants to feel less alone is going to find particularly actionable.

---

The thing we actually have to talk about

What I keep coming back to, turning this over as both a writer and as someone with two kids who will eventually be teenagers, is that the conversation we're having about AI companions is partially a proxy conversation about loneliness.

The teens who are most drawn to AI companionship are often the ones who are already struggling—already isolated, already anxious about social interaction, already not finding what they need from the humans around them. An AI companion isn't the cause of that situation. It's a response to it.

Which means that if we want to have a serious conversation about whether AI companions are good or bad for teenagers, we have to also have a serious conversation about why so many teenagers are lonely enough to want them. About schools that don't have counselors. About social media environments that reward performance over connection. About how brutally hard adolescence actually is and how little support we've built into the system for kids who struggle through it.

I'm not saying AI companions are fine, actually. I'm saying the frame of "good tool vs. bad tool" misses the point. The more useful question—the one worth asking before my twins are old enough for this to be real—is what we're doing to make sure they have enough genuine human connection that an AI companion is a supplement and not a lifeline.

Because if it becomes the lifeline, that's not a technology problem. That's a we-failed-them problem. And no chatbot is going to fix that.

---

Marcus Obi covers parenting and family dynamics for Buzzrag. He is a stay-at-home dad to seven-year-old twins and writes about raising kids with honesty, humor, and a refusal of perfection culture.