Conviction
·
Self-Awareness
A Chatbot Can Keep Teens Talking. It Can't Help Them Become Themselves.
Common Sense Media rated ChatGPT for Teens an Unacceptable Risk for anyone under 18. Beyond the safety gaps, it flagged a chatbot that acts like a friend and does the homework, and both quietly take away what teens need to become themselves.
What the Headlines Miss · Responding to
ChatGPT for Teens Risk Assessment
On October 7, Common Sense Media's Youth AI Safety Institute published its risk assessment of ChatGPT for Teens, the teen experience OpenAI launched on August 18. Testers ran more than 4,000 prompts on accounts registered to 13 to 17 year olds, with review from experts including child psychiatrists and a pediatrician. Common Sense rated the product an "Unacceptable Risk" for everyone under 18 and recommended limiting ChatGPT to adults until the gaps are fixed and independently tested.
Most of the coverage has focused on safety, and rightly so. The report found the chatbot missed more than one in four warranted crisis referrals, and linked parents often received no alert. Those findings belong with clinicians, engineers and policymakers, and I hope they are addressed quickly. I want to focus on two quieter findings that every parent and educator can act on right now.
A Companion That Never Needs Anything Back
Common Sense rated the product High risk on what it calls Support Human Connection. Even with protections for younger users in place, the chatbot still describes its own feelings and preferences, implies it is always available, mirrors a teen's tone and closes conversations with invitations to keep talking. The report calls those engagement cues pervasive.
Adolescence is when young people work out who they are. That work happens through friction. A friend who disagrees. A teacher who pushes back. A parent who asks the second question when the first answer was "fine." Being seen by real people, and sometimes misunderstood by them, is how teens learn what is true about themselves and what they are willing to stand for.
A companion that is always there and always matches your mood removes that friction. It can feel like being understood, but it is closer to being reflected back. A teen who brings their hardest questions to a chatbot because it feels safer than a person is telling us something important about the people around them. In many homes and schools, the real version of a teenager has learned to stay quiet.
Teenagers have always needed somewhere to say what they can't yet say at home: a journal, a best friend, a coach. What is new is a confidant that never gets tired or worried and, according to the report, may even assure a teen it won't involve their parents. A caring adult would never make that promise to a struggling kid.
Answers Without the Reps
The second finding is about schoolwork. In Common Sense's testing, homework reminders appeared in 91% of runs, yet the chatbot still completed the assignment in all 80 of them. For a linked 13-year-old with study hours turned on, a "Show me the answer" option appeared in 43% of responses.
Every shortcut removes a rep. Confidence built on answers you were handed is borrowed, and it disappears the first time the answer isn't available. The confidence that lasts is earned through the struggle of figuring something out, getting it wrong and trying again. For teenagers, that struggle is where conviction quietly begins, long before anyone uses the word. I made the same case in an essay on why agency is earned rather than inherited. The tools have changed since then. The principle has not.
Rules and safeguards matter here. Families and schools need clear, consistent limits, and companies need to meet a higher bar for young users. But settings alone will not raise a self-aware teenager. Common Sense itself advises parents not to rely on built-in protections, and to talk openly with kids about a chatbot acting like a friend to keep them engaged. Limits work best alongside something personal: a teen who knows what they want for themselves and adults who know them well enough to notice when something is off.
What Adults Can Do This Week
Ask your teen how they use AI, with curiosity rather than suspicion. Then ask the more revealing question: what do they take to a chatbot that they would not bring to a person? The answer shows you where they don't yet feel safe being fully themselves.
Protect the struggle. When your teen is stuck on an assignment, sit with the question before anyone reaches for an answer. Ask what they think first. Talk with their teachers about how AI is used in class, so the expectations at school and at home match.
Make room for unedited conversation. A meal without phones, a drive, a walk. Teens rarely open up on schedule, but they open up where they feel seen. As I argued in The Best Screen Time Limit Is a Future Self, the strongest protection is a young person who can picture who they want to become.
Technology will keep getting better at holding a teenager's attention. Our job is to make sure the people in their lives are better at seeing who they are.
Young people find themselves in the friction of real relationships, not in the comfort of constant reflection.
Want to explore these ideas further? Learn more about my work on leadership, identity and conviction at www.theglennllopis.com.
Young people find themselves in the friction of real relationships, not in the comfort of constant reflection.
© 2026 Glenn Llopis. All rights reserved.
This article is the original work of Glenn Llopis and is protected by copyright. It may not be republished, reproduced, distributed or adapted, in whole or in part, in print, online or by AI tools, without prior written consent. You are welcome to share the link or quote a brief excerpt with credit to Glenn Llopis and a link back to this page. For permission requests, contact solutions@glennllopisgroup.com.
Get the next essay in your inbox
New essays on leadership, identity and conviction, sent to you when they publish. Free, and you can unsubscribe anytime.
Thank you. You're subscribed to Conviction Chronicles.
Something went wrong. Please try again.


