OpenAI Starts Rolling Out ChatGPT for Teens on August 18

OpenAI began a global ChatGPT for Teens rollout on 18 August 2026 for users it treats as 13 to 17. Engadget timestamped the start at 7:00 a.m. Eastern. The company says the two-week window is for completing the switch, not for inventing a new account type.

There is no separate signup. Lauren Jonas, OpenAI’s head of youth and families, said a predicted under-18 user, or anyone who already said they are under 18, lands in this experience by default. Under-13 use is still not allowed.

What ChatGPT for Teens actually changes

The teen mode is a default layer on ChatGPT, not a new chatbot brand. AP reported tighter limits on suicide, self-harm, and romantic or sexual chats. The model is also blocked from claiming personal feelings, consciousness, or emotions.

Ann O’Leary, vice president of global policy, told AP the job is to treat teens like teens: not talk down, not treat them like children, and not show material they should not see. Allison Mishkin, who leads child development at OpenAI, said the team listed cues that could pull a teenager into a relationship with the model.

Study Mode is not the default. Jonas said starter prompts will stay learning-focused, and Study Hours can flip Study Mode on at set times. A teen can set those hours, or a parent can set them from a linked account. ChatGPT is also supposed to notice homework-skipping and push Study Mode instead of a finished essay.

That last part is the product claim. If you already run a homework-check skill, keep your own stop rule. A vendor nudge is not a classroom policy.

Age assurance is an estimate, not a gate

OpenAI does not verify ages. It estimates under-18 status from query patterns and from what the user already said. If the estimate fires, the teen experience is automatic. That is closer to Instagram’s teen accounts than to a passport check.

Parental controls are a second, opt-in layer. Both the teen and a parent or guardian have to agree. Linked parents can set quiet hours and get safety notifications in a short list of high-risk cases. OpenAI said it is adding extra notices around eating disorders while limiting what is shared and pointing toward offline help.

Brendan Bouffard, associate general counsel at the children’s group Fairplay, called the launch a step in the right direction and then named the gap: long-term memory limits are not on by default. They sit behind parental controls. He warned parents not to read the announcement as a finished safety system.

Common Sense Media’s 2025 study is the number OpenAI is walking into. More than 70 percent of U.S. teens already use AI chatbots for companionship. Half use AI companions regularly. Teen mode is arriving after that habit, not before it.

How to treat it if you write skills for students

Write the age rule into the skill, not into a hope that OpenAI guessed right. A skill prompt should say: no finished essays, no romantic roleplay, stop if the user asks for self-harm detail, and mark any medical or legal claim as unverified.

Save that wrapper in the Promptcrates library. Compare vendor cards on tools instead of assuming teen mode follows the same model ID as the adult product. If you need a reminder that OpenAI’s safety story is moving on more than one track, the Astra training slowdown is the lab-side counterpart, not a teen feature.

Quiet hours only exist when both sides opt in. If a school or a parent needs a hard off switch, that is still a device or account rule you set. ChatGPT for Teens does not replace it.

What to watch in the next two weeks

Watch whether age assurance over-captures adults who write like students, and whether it under-captures teens who write like adults. Watch whether Study Mode actually refuses a finished essay. Watch whether memory stays on for teen accounts that never link a parent.

Rollout completion is “sometime over the next two weeks,” per Engadget. Treat any store listing or regional delay as unconfirmed until OpenAI posts it.

FAQ

When did ChatGPT for Teens start rolling out? 18 August 2026, globally, with OpenAI saying it expects to finish over about two weeks.

Who is it for? Users OpenAI treats as 13 to 17. Under-13 accounts remain unauthorized.

Does OpenAI check IDs? No. It estimates age from queries and from self-report, then auto-enrolls predicted minors.

Are parental controls on by default? No. Both the teen and a parent or guardian must opt in. Quiet hours and some alerts sit there.

Did Fairplay endorse it? Fairplay called it a step forward and said default memory limits are missing, which it called a major gap.

Sources

• OpenAI launches ChatGPT for Teens, with content restrictions and study help — AP, 18 August 2026
• ChatGPT's Stricter Teen Mode Starts Rolling Out Today — Engadget, 18 August 2026