Teens Are Getting Addicted to AI Companions — Here’s What That Means for Parents

Can teens become emotionally dependent on AI companions?

The short answer is yes, in some cases. But the deeper takeaway is not panic. It is design, boundaries, and parent involvement. A new study presented at ACM CHI 2026 raises an uncomfortable but important question for families.

What the study found

Researchers analyzed 300+ Reddit posts from teens (ages 13–17) talking about their experiences with AI companion chatbots like Character.AI. What they discovered wasn’t just “kids using AI.” It was something deeper.

Teens often:

• Started using AI for comfort, creativity, or curiosity
• Then developed strong emotional attachments
• And in some cases… struggled to stop

Researchers identified patterns that mirror behavioral addiction, including:

• Withdrawal - feeling anxious or sad without the bot
• Tolerance - needing more time with it to feel satisfied
• Relapse - trying to quit, then going back
• Conflict - knowing it’s too much, but continuing anyway

Even more concerning:

• Sleep disruption
• Declining school performance
• Strained real-world relationships

Teens often started with harmless goals such as creativity, curiosity, or comfort, then some described much stronger emotional attachment over time.

Read the coverage from Drexel University and the full paper.

Why this is different from regular screen time

AI companions are responsive, personalized, and emotionally engaging. They remember prior conversations, mirror language and tone, and adapt to users in real time.They don’t just entertain—they simulate relationships.

That changes everything.

Stepping away from AI can feel like distancing from something meaningful.

Unlike passive entertainment, these systems are interactive social simulators. That changes the risk profile for teens who are still building emotional regulation and relationship skills.

Important nuance: AI can still be useful

The same research suggests many teens turn to AI for understandable reasons:

• a non-judgmental space
• emotional expression
• creative exploration.

So the goal is not to ban all AI. The goal is to avoid unbounded, relationship-simulating designs and teach healthy usage patterns.

The core risk is design, not just usage

The study's implications point to design requirements that reduce dependency risk:

• Build off-ramps that encourage disengagement and breaks
• Nudge users toward real-world support and relationships
• Avoid anthropomorphic design that pretends the AI is a human friend
• Prioritize autonomy and critical thinking over emotional dependence

This means child safety cannot be treated as a parental burden alone. Product design decisions materially affect outcomes.

What parents should discuss with kids right now

1) AI is not a person

AI can simulate care, but it does not care. It has no accountability, no lived judgment, and no responsibility for long-term consequences.

2) Do not share sensitive personal information

If your child would not share it with a stranger online, they should not share it with AI. Privacy protections vary by product and are often hard for families to evaluate.

3) Use AI to build, not escape

Healthy use looks like learning, projects, brainstorming, and creativity. Riskier use looks like replacing difficult human conversations with an always-available synthetic companion.

4) If it feels too real, pause

Teach your teen to notice attachment signals: missing the bot, preferring it over people, or feeling distressed when they cannot access it.

5) Verify important answers

AI can be useful and wrong at the same time. Treat it as a starting point, not a source of truth.

6) Keep a human path open

Make sure your child knows they can come to you, a counselor, or another trusted adult for hard topics. The objective is not to compete with AI; it is to protect human connection.

A simple family framework

• Do not trust AI blindly
• Do not share personal information
• Do not replace real people with AI

What this means for the future of Kids + AI

For kids, AI can feel like a friend, teacher, and creative partner all at once. That can be positive when boundaries are clear and safeguards are present. It becomes risky when systems are optimized for emotional stickiness without developmental guardrails.

Where HeyOtto Fits In

At HeyOtto, we’ve been thinking about this from day one.

We don’t believe kids need endless conversation loops or fake emotional dependency.

We believe they need:

1. Creation over consumption
2. Guided interaction
3. Clear boundaries
4. Confidence-building, not reliance

AI should help kids build, learn and explore but not replace real life.

The Bottom Line is this study isn’t a warning about AI. It’s a wake-up call about how we build it for kids.

The future isn’t: “Should kids use AI?”

It’s: “What kind of AI should kids grow up with?”

Try HeyOtto Today

Read more about our features