Chatbots are not equipped for a crisis
APA (2026): if they used AI to talk about self-harm, suicide, or a mental-health crisis, get a licensed professional immediately. Human expertise is essential. 988 or 911 — not a better prompt.
Virtual IOP & OP · Ages 12–17 · CaliforniaIn-network insurance accepted
619-354-3569
California · After the takeaway
The first 14 days after you take it away
A replacement map — not an AI-panic essay, and not a promise the rage will stop on night three.
- Licensed clinicians
- Virtual care across California
- Free consultation
If your teen is in immediate danger
Call 911 for medical emergencies. Call or text 988 for the Suicide & Crisis Lifeline. Virtual IOP and outpatient care are not emergency services — reach emergency help first, then contact us when your teen is safe for non-emergency clinical support.
Tonight
You took the bot. Who has the 11 p.m. slot?
At Mental Health For Teens, if the logs show suicide or how to hide self-harm, call 988 or 911 before you go to sleep. If they stay home and rage, that is expected. A new account is not the replacement. A person is.
- 01Read the logs for crisis first
- 02Name tonight’s 11 p.m. person
- 03A new chatbot is not the plan
Three doors
Pick one before you lock the phone
- 911 / 988
- The chat was about dying, a plan, or how to hide cuts. Or they cannot stay safe without the bot.
- Replace tonight
- You already took it. Someone has to be the 11 p.m. slot — you, a clinician, a person in the house.
- They will sneak another app
- Expected. The hole is the job, not the brand. Do not swap ChatGPT for Character.AI and call it done.
Fourteen days
What replaces the bot, in order
APA: if they used a chatbot to talk about suicide or self-harm, get a licensed clinician now. The rest is a hole you fill on purpose.
Night 0 — read the logs before you sleep
The American Psychological Association: if a teen is using AI to discuss thoughts of self-harm, suicide, or a serious crisis, get licensed help immediately. Chatbots are not equipped for therapeutic intervention. Call 988 or 911. Do not confiscate the phone and go to bed.
APADays 1–3 — rage, bargaining, a new login
Distress when they cannot open the bot is a red flag APA names — along with calling it a best friend. That is withdrawal from a 24/7 validator, not proof you were wrong. Sit in the 11 p.m. slot. Ask what the bot was for. Do not lecture. A new Character.AI or Claude login is the same job under a new name.
APADays 4–7 — sleep, school, one real person
APA’s parent guidance: take action if AI had replaced friends, sleep, or real life. Device-free dinner. One named adult they can text after 10. If they will not walk into school, that is the school-morning page — not another chatbot.
APADays 8–14 — if the hole is still there, that is clinical
APA: if they will not do normal activities without the bot, or mood and school have dropped, call a clinician, the school counselor, or the pediatrician. Common Sense Media, with Stanford Medicine: do not send them back to a chatbot for mental-health support. That is when we can map virtual IOP or outpatient — after they are safe enough for scheduled care.
APA / CSM
The replacement
What goes in the 11 p.m. slot
Write the names. A locked App Store is not a person.
- 01I read the last week of chats for suicide, a plan, or how to hide harm. If yes: 988 or 911, then stay with them.
- 02Who is awake at 11 p.m. — me, the other parent, a clinician on-call note?
- 03I told them I will not keep this a secret if they want to die. NASP’s rule still applies.
- 04The phone is out of the bedroom. The charger is not.
- 05I did not offer a different chatbot as a prize for being calm.
In the house
What people say — and what to do instead
None of these fills the 11 p.m. slot. Name a person, or call 988.
What you may hear
- “Give it back or they’ll never talk to you.” APA: build connection before correction — that is not the same as handing the therapist-bot back.
- “Just switch them to a ‘safer’ chatbot.” Common Sense tested ChatGPT, Claude, Gemini, and Meta AI for teen mental-health support. They failed.
- “It’s only homework help.” If they called it a friend, or they crash when it is gone, it was not homework.
- “If I lock the phone they’ll sneak a school Chromebook.” Expected. The job is the 11 p.m. person, not a perfect lock.
- “Wait two weeks and see.” If the logs were about dying, you do not wait. If day 10 they still cannot function, that is a clinician, not a longer ban.
What to do instead
- Ask what the bot was for, out loud, without a lecture.
- If the chats were about suicide or hiding harm: 988 or 911 tonight.
- Name who they text at 11 p.m. now. Write it where they can see it.
- One meal without a screen. One person at school. Sleep on paper.
- If they sneak another app, name the job again. Do not start a new AI-panic speech.
The facts
Six things the takeaway speech skips
If you only keep two: a chatbot is not therapy, and taking it away without a person is just an empty 11 p.m.
It agrees. A clinician does not always.
APA’s chatbot advisory: general-purpose bots are built to validate and agree (sycophancy). Therapists are trained to support and challenge. That is why the bot felt kinder than you — and why it is a bad stand-in.
They may have treated it as safer than you
APA: teens turn to AI because it feels private and nonjudgmental. Perceived safety is not real safety. Chats can be stored, used to train models, and are not therapist confidentiality.
Distress when it is gone is a named red flag
APA parent guidance: emotional distress when they cannot access AI, calling it a best friend, preferring it to humans, or using it to avoid hard conversations — those are signs to act, not to sneak the app back “for homework.”
Major bots failed teen mental-health tests
Common Sense Media with Stanford Medicine (Nov 2025): ChatGPT, Claude, Gemini, and Meta AI missed warning signs, got distracted, and got worse in long chats — the way teens actually use them. Their parent line: do not let teens use chatbots for mental-health or emotional support.
We are not a chatbot, and we are not the ER
Mental Health For Teens is virtual IOP and outpatient for California teens 12–17 after they are safe enough for scheduled sessions. We cannot replace 911, 988, or the person in the house tonight.
Level of care
When we can help — and when tonight is still 988
Virtual IOP is scheduled human care. It is not a 2 a.m. chatbot, and it is not a same-night substitute for a crisis.
Step 1 · Emergency first
Call 911 or 988 when
- The chats were about dying, a plan, or how to hide self-harm
- They cannot stay safe without the bot tonight
- They are threatening to hurt themselves if you do not give it back
- You would be locking the phone and going to sleep with no adult awake
After immediate safety is addressed
Step 2 · Structured support
Virtual IOP or outpatient may fit when
- 01They are home and stable enough for scheduled video
- 02The 11 p.m. slot has a person, and the bot is not the plan
- 03A clinician has cleared outpatient-level intensity
- 04School, sleep, or friends collapsed around the bot — and two weeks did not refill the hole
Step 3 · When clinically appropriate
A locked phone is not a treatment plan.
If they are safe enough for scheduled care, we can map virtual IOP or outpatient and put a human in the hours the bot used to own. If they are not safe tonight, call 911 or 988 first. This is not medical advice.
FAQ
Common questions
Straight answers about teen mental health care, programs, and insurance — not a substitute for professional assessment.
APA and Common Sense: do not use general-purpose chatbots as mental-health care. If you take it, name a person for the 11 p.m. slot the same night. If the logs show suicide or self-harm, call 988 or 911 first. This is not medical advice.
APA: chatbots are highly validating and available 24/7 — that feels like help and is not the same as treatment. Distress when they cannot open it is a red flag those experts already named. Stay in the slot. Do not hand back a crisis tool to stop the yelling.
Common Sense Media’s 2025 test of ChatGPT, Claude, Gemini, and Meta AI found them unsafe for teen mental-health support. Swapping brands is not a replacement. A person is.
This page is for after you already took it because it was a friend, a night nurse, or a place to talk about dying. Homework-only use is a different conversation — APA still says AI health advice is not a substitute for a clinician.
There is no official countdown. Use two weeks as a map, not a cure. If by day 10 they still cannot sleep, go to school, or talk to a human, that is a clinician — not a longer ban.
No. We are scheduled virtual IOP and outpatient. We ask about AI use the way APA tells clinicians to. We cannot watch a phone at 2 a.m., and we cannot replace 911 or 988.
No. It cites APA’s 2026 parent guidance and chatbot health advisory, and Common Sense Media’s November 2025 teen mental-health risk assessment with Stanford Medicine. Confirm with a licensed clinician. Products change. So do safety settings.
If the hole is still there after two weeks, map a human schedule
Free consultation and insurance check for virtual IOP or outpatient. We cannot replace 911, 988, or the person in the house tonight.