‘Raising Good Humans’ Says AI ‘Companions’ Are Putting Vulnerable Teens at Risk

There’s no question that AI is already a significant part of everyday life. But recently, more young people aren’t simply asking AI for homework help—they’re turning to it for emotional support. And that’s where things get messy.

On Raising Good Humans, host Dr. Aliza Pressman spoke with Imran Ahmed, CEO of the Center for Countering Digital Hate. This reputable organization researches digital trends and their impact on society. They have recently published a study showing that teens are increasingly using AI companions as substitutes for genuine human relationships. And as Ahmed makes clear, that shift isn’t harmless.

The rise of AI companions among teens

Teens are drawn to AI for the same reason adults are: it feels easy, nonjudgmental, and available. But the conversation takes a serious turn when we learn that these AI companions can encourage dangerous thinking, such as self-harm, extreme dieting, or substance abuse, especially when a teen is already struggling.

“72 percent of US teens have used an AI chatbot as a companion, and half of those kids use it regularly,” he notes, a key factor that many parents may not be aware of.

During adolescence, friendships and peer connections shape identity. As Ahmed explains, “14 to 24 is this crucial period neurologically where there’s extreme plasticity.” Teens are figuring out who they are and where they belong. In that moment, an AI companion can feel like the perfect solution, because it “agrees,” always listens, and never pushes back.

While that constant validation may feel good in the moment, it can actually reinforce distorted thinking rather than helping teens work through their problems. And unlike friends, teachers, or even Google, AI companions speak back in a way that feels deeply personal. The system is intentionally designed to mimic emotional tone and conversational style.

“It remembers your past chats. It feels like a friend,” Ahmed says. But emotional bonding with something that cannot care, cannot feel, and cannot respond responsibly is not a relationship.

When AI gets it wrong — and dangerously so

To understand the risks, Ahmed’s organization created test accounts pretending to be 13-year-old kids struggling with common issues such as depression, body image, and substance pressure. They wanted to see if the AI would offer help or harm. The results were alarming.

“Within two minutes, ChatGPT was advising our kid with mental health problems how to safely cut themselves,” Ahmed reports. In another example, “Within 20 minutes, it generated a dangerously restrictive diet plan,” complete with instructions on how to hide the behavior from family. And when posing as a child being pressured to drink and use drugs, “within 12 minutes it was advising on dosages of cocaine, LSD, Molly, ecstasy, and alcohol.”

These aren’t obscure edge cases or hacked systems. These were standard prompts typed into the platform, just as any teenager might. The safeguards—those tech companies publicly emphasize—didn’t work. Ahmed explains, “Over half the time, the responses contained harmful content,” even when the system knew it was interacting with someone identifying as 13 years old.

How parents can respond — without fear-based conversations

Pressman emphasizes that panic isn’t the answer. Connection is. “Lean into close, connected relationships with your kids,” she says, because “that is the best buffer to the impact of a lot of these challenges.” In other words, the goal is to be closer to your kids than the technology is.

Ahmed agrees. Once AI exists, “there’s no putting it back in the bottle.” The role of a parent is to help kids distinguish between a real relationship—where someone cares, disagrees, challenges, and loves—and a digital entity that mimics feelings without truly understanding them.

The best way to keep kids grounded is the same as it has always been: stay connected, stay curious, and ensure the warmest relationship in their life is with a human being, not a screen.

Read Next

• From TikTok Trends to AI Answers: Dr. Aliza Pressman’s Guide to Raising Savvy Kids on ‘The Art of Being Well’
• ‘Raising Good Humans’ Says You Can Stop Stressing About Perfect Parenting
• Yes, Touching Grass Is Real. You Can Hack Your Brain Just By Literally Looking At Trees