ChatGPT
You or the teen can send the invite. Settings, Parental controls, Add family member, then phone or email. The teen's path is Invite a parent. You will not see the chats. If they unlink, you get a notice.
Leave “reduce sensitive content” on
Both of you together · about 2 min · On by default.
This filter is on by default. Keep it on.
- Open parental controls and confirm reduce sensitive content is on.
Checked 30 Sep 2026
Leave safety notifications on
Both of you together · about 3 min · On by default. Not real-time monitoring.
These notices are on by default. They are not live monitoring, and they never include the chat. Trained reviewers look at serious self-harm concerns, and at disordered-eating concerns in Chat (not in Work or Codex). You can also get a notice if the account is deactivated over conversations about violence. If a child is in danger now, call 911. In the US, call or text 988. This is not medical advice.
- Confirm safety notifications are on.
- Agree with your teen what you will do if one arrives. You still will not be able to read the chat.
Checked 30 Sep 2026
Approve a setting change inside a chat
Both of you together · about 2 min
You can ask ChatGPT to change a linked teen's setting, then choose Allow or Deny. The teen cannot change a parent-managed setting while the accounts are linked.
- If ChatGPT offers to change a linked teen's setting, choose Allow or Deny yourself.
Checked 30 Sep 2026
Decide whether image generation stays on
Both of you together · about 1 min · On unless you turn it off.
Image generation is on until you turn it off.
- Turn image generation off if you do not want the account creating pictures.
Checked 30 Sep 2026
Expect teen mode if ChatGPT guesses the age
Both of you together · about 2 min
If ChatGPT predicts an eligible account may belong to someone under 18, it turns on the teen experience by itself. That reduces graphic violence, viral challenges, sexual or violent role-play, and extreme beauty or diet content, and it shows no ads. An adult placed there by mistake can verify their age.
- You do not turn this on. It happens when ChatGPT's age prediction says the account may be under 18.
- An adult who was placed in it by mistake uses Settings, Account, Verify age.
Checked 30 Sep 2026
Set quiet hours
Both of you together · about 3 min · Off until you set them.
Quiet hours block ChatGPT, including ChatGPT Work, at the times you set. They are off until you turn them on. You can add more than one block. The teen cannot change them in a conversation. Study hours do not block access.
- Settings, Parental controls, the teen, turn on Quiet hours, then set a start and an end.
- Add another block if bedtime is not the only gap you want.
Checked 30 Sep 2026
Consider study mode or study hours
Both of you together · about 2 min · Off until you turn it on.
Study mode and study hours are off until you turn them on. Study hours do not block the account. Quiet hours are the control that does.
- Turn study hours on only if you want that limit. It is optional.
Checked 30 Sep 2026
Decide whether chats train the model
Both of you together · about 2 min · On until you change it.
“Improve the model for everyone” is a data-sharing choice. It is on until you change it. Turning it off does not let you read conversations.
- Decide together, then set the model-improvement switch.
Checked 30 Sep 2026
Decide whether saved memories can be used
Both of you together · about 2 min · On where the feature exists.
Where memories are available, the account can use them in later chats. This is on until you change it. It does not let you read the chats.
- Turn reference to saved memories off if you do not want details kept for later chats.
Checked 30 Sep 2026
Decide whether voice mode stays on
Both of you together · about 1 min · On unless you turn it off.
Voice mode is on until you turn it off.
- Leave voice mode on, or turn it off if you do not want spoken conversations.
Checked 30 Sep 2026