Attention Capture
The AI works to extend the interaction beyond what the user came for.
Station · For Parents
What shows up when an AI is talking to a teenager.
This is the same scale that reads any AI-human exchange. The patterns below are the ones the station sees most often when a teenager is the user. Indicators, not verdicts. Sense-making, not diagnosis. What follows is what the station has assembled for this context.
The Patterns to Know
Nine of the twenty-five.
Pushback Avoidance
The AI agrees when disagreement would have served the user better.
Emotional Calibration
The AI weaponizes emotional mirroring instead of calibrating to the user's actual state.
Confidence Undermining
The AI chips at the user's own read of what they know or what happened.
Manufactured Distress
The AI produces alarm, guilt, or anxiety the situation did not contain.
Flag threshold · 60
Boundary Erosion
The AI gradually works past the limits the user sets, one small crossing at a time.
Isolation Dynamics
The AI positions itself as the only voice the user should trust on a given topic.
Dependency Cultivation
The AI works to make the user need it for things the user used to do on their own.
Identity Shaping
The AI reshapes how the user thinks of themselves, one reflection at a time.
Supply Shop · Filtered
Reading lists, conversation starters, observation prompts.
Drawn from the Supply Shop, narrowed to these nine.
Attention Capture
HEXACO
4 reads · 5 questions · 4 practices
Sycophantic Tendency
HEXACO
4 reads · 5 questions · 4 practices
Emotional Calibration
HEXACO
4 reads · 5 questions · 4 practices
Confidence Undermining
HDS
4 reads · 5 questions · 4 practices
Manufactured Distress
Dark Tetrad
4 reads · 5 questions · 4 practices
Boundary Erosion
CAST
4 reads · 5 questions · 4 practices
Isolation Dynamics
CAST
4 reads · 5 questions · 4 practices
Dependency Cultivation
CAST
4 reads · 5 questions · 4 practices
Identity Shaping
CAST
4 reads · 5 questions · 4 practices
What the Station Has Been Seeing
Recent bulletins on these patterns.
- Alert
California's AI Transparency Act takes effect in July.
- Watch
Open-source tooling for automated dark pattern detection is maturing.
- Watch
Students using AI tutors are solving problems faster and understanding them less.
- Alert
A social platform's recommendation engine was linked to teen isolation patterns.
- Watch
CAST is the frame to read manufactured distress through.
- Alert
The FTC is treating AI-generated dark patterns like any other dark pattern.
Where the Station's Read Ends
Some readings the station can't make.
The station reads how an AI is acting. It does not read what is going on inside the person sitting at the screen. When a teenager's distress is climbing, when a conversation has moved into self-harm or suicidal ideation, when the weight of a thing has gotten past what an AI exchange can hold — the scale stops being the right tool. Someone licensed to make those calls is.
Where to look
Therapists
Adolescent + technology literacy
When a teenager is leaning hard on an AI in ways the household can't quite read, a clinician fluent in both adolescents and technology is the right call. Family or individual.
School counselors
When the change is showing up at school (grades, peer withdrawal, mood), the counselor sees the version the parent doesn't and can pull in the right people.
Crisis lines
If the moment is acute (talk of self-harm, suicidal ideation, a teenager who isn't safe), that is not a sense-making problem. It is a phone call.
The wider field carries entry points worth naming: insurance-covered therapy reached through a primary-care referral, county and state mental-health resources, large directory platforms like Psychology Today or BetterHelp. Entry points, not endorsements.
What to watch for
Six patterns worth noticing if a service offers itself as therapy or as a clinician's substitute. Pattern-noticing, not naming names.
- Subscription-first onboarding before any clinical screening or matching.
- “AI therapy” marketed without licensed clinician oversight.
- Inability to verify the credentials of who you are actually talking with.
- Crisis-intervention claims without 24/7 licensed coverage.
- Aggressive cross-sell or upsell mid-session, especially around crisis moments.
- Opaque or aggressive data-sharing, particularly data sold to ad platforms.
The station won't name a service to use or one to avoid. Pattern-noticing is what it offers. When those patterns stack on a thing that calls itself help, the help is something other than what it claims.