ChatGPT for Teens: OpenAI’s Aug 2026 teen mode, study and limits
OpenAI launched ChatGPT for Teens. Official text: self-reported ages 13–17 or an under-18 estimate auto-enrol. Parents cannot read chats. This is a product note, not an efficacy trial.
On 18 August 2026 OpenAI framed teen ChatGPT as its own experience, not another general model drop. The name is ChatGPT for Teens.
The official post says anyone the system estimates is under 18, or who states an age of 13 to 17, is placed in automatically. Help pages add account age, verified age, and age prediction. Lines that match those pages are marked official. BBC, TechCrunch, and Engadget sit in a separate box.
What the official pages print
| Item | Who said it | Claim |
|---|---|---|
| Auto-enrol | Official / help | Under-18 estimate or ages 13–17 |
| Parents read chats | Help | No |
| Ads | Help | Not shown on prediction-triggered teen accounts |
| Under 13 | BBC | Not allowed (reporting, not the post’s wording) |
- Four commitments
- Official: put teen safety first, encourage real-world support, treat teens like teens, and be transparent about system behavior. Earlier work: Parental Controls, the Under-18 Model Spec, the Teen Safety Blueprint, and age prediction.
- Learning
- Help: Study Mode can give hints, steps, follow-ups, and knowledge checks; homework reminders may suggest Study Mode; Study Hours start eligible new chats in Study Mode.
- Parental controls
- Optional. Study Hours do not block ChatGPT. Quiet Hours can limit access on a schedule. Safety notices share only what is needed to support safety.
- Age prediction
- Help: rolling out globally; the EU follows in coming weeks. In Italy, after a verify prompt, there are 60 days before some features may shut off.
How an account enters and leaves
- 1
No toggle required
Help: eligible accounts get it automatically and stay active. Age verification is not required to keep the protections.
- 2
Adults verify to exit
Help: if you are 18+ and think it is a miss, verify age to drop teen protections. Persona handles verification.
- 3
Parents link separately
Help: parental controls need a linked account. AP says both sides must opt in. Quiet Hours and safety notices sit on that link.
- 4
Still rolling
Help: if teen features are missing, the account may not be in the wave yet. Australia full availability is expected 8 September.
What is still unmeasured
No efficacy trial
OpenAI did not publish an independent test that cheating or dependence fell. Fairplay told AP that an announcement is not the same as safety in place.
Memory defaults
AP quotes Fairplay: long-term memory is not off by default and needs parental controls — a gap in that group’s view. Help does not spell a memory default in the same section.
Workarounds
TechCrunch notes teens may route around controls. OpenAI gave no bypass rate.
Limits
No full table of country-by-country finish dates. No accuracy figure for parent notices. No grade-impact study for Study Mode. Next steps are more features and published learnings — without a date.
# Sources
# OpenAI, 2026-08-18: Introducing ChatGPT for Teens
# OpenAI Help: ChatGPT for Teens; Age prediction in ChatGPT
# BBC / TechCrunch / Engadget, 2026-08-18 (labeled where used)Can every teen see it now?
Help says the rollout began 18 August; it was not a same-day global flip. Australia full availability is expected 8 September. Missing UI means the account may still be waiting.
Can a parent read the chats?
Help says no reading and no monitoring. Safety notices share only what is needed to support safety.
Is this a new model?
Help calls it the same capable ChatGPT plus age-appropriate protections and optional parental controls. The official post does not name a separate model.