Is it safe to upload your child's photo to ChatGPT?

Published November 3, 2026

Uploading a photo of your child to ChatGPT is not dangerous in itself, but it is not private by default. Two separate things happen: the photo is stored with your conversation, and on personal plans (Free, Plus and Pro) the conversation may be used to train OpenAI's models unless you turn that off. If you do upload, turn off "Improve the model for everyone", use a Temporary Chat, skip the thumbs up and down buttons in that chat, and crop out anything that shows where your child lives or goes to school.

The settings below are as described in OpenAI's help center on October 25, 2026. They change; check OpenAI's data controls page before relying on them.

What the settings actually do

Setting	What it does	What it does not do
Improve the model for everyone: on (the default on Free, Plus and Pro)	Your conversations may be used to train OpenAI's models
Improve the model for everyone: off	New conversations are not used to train OpenAI's models	Does not delete or hide chats you already have; they stay in your history
Temporary Chat	Not shown in your history, does not create or update memories, not used to improve the models	Not deleted immediately; OpenAI says these chats may be kept for up to 30 days for safety purposes
Thumbs up or down on a reply	Sends feedback	Even after you opt out of training, OpenAI says the entire conversation tied to that feedback may be used to train its models
Business, Enterprise, Edu, or the API	Not used for training by default	Your organization's retention policies still apply

The feedback row is the one most people miss: after you opt out, OpenAI says that if you give feedback on a reply, the entire conversation associated with that feedback may be used to train its models. That includes a photo you uploaded in it.

Is it safe? A security view

A practical way to think about it: where does a copy of the photo now exist, and for how long?

A copy lives with the provider. Whatever the training setting, the provider stores your conversation in your history until you delete it, and Temporary Chats for up to 30 days. That copy is protected by the provider's security, not yours.
Training is a separate question from storage. Opting out of training keeps your photo out of future model training. It does not make the conversation disappear.
The file may carry more than the picture. Photos and videos can contain GPS coordinates and device details. We have not tested what each assistant does with that data, so remove location before you upload. Here is what one phone video contains.
Other assistants work differently. Gemini, Claude, Copilot and others each have their own settings and defaults. Check each one rather than assuming they match.

Habits that keep it safe

Ask whether you need the photo at all. Many questions ("is this rash normal", "what should we name this outfit") can be described in words.
Use a Temporary Chat for anything involving your child, and do not save it.
Turn off "Improve the model for everyone" once, in Settings, then Data controls. It applies across web and mobile when you are signed in.
Skip the thumbs up and down in chats that include your child.
Crop and strip before uploading. Remove faces you do not need, school logos, house numbers and street signs, and share without location (on iPhone: Share, Options, turn off Location).
Delete the chat when you are done, and remember that deleting is separate from opting out of training.
For medical concerns, ask your pediatrician. A doctor's office is set up to handle your child's health information; a chatbot is not a substitute.

Where WholeFamily stands

We get asked whether WholeFamily sends photos to AI. It does not. Two features send text to an AI model, translation when you ask for one and the daily email digest, and never photos or videos. That text is processed only to produce the translation or summary, under terms that prohibit using it to train models. The privacy policy lists exactly what is sent, and this post explains how we encrypt your photos.

The short version

Not private by default on personal ChatGPT plans: training is on until you turn it off.
Turn off "Improve the model for everyone" or use a Temporary Chat, and skip the feedback buttons.
Opting out of training is not deleting. Delete chats you do not want kept.
Strip location and identifying details before you upload anything of your child.