Skip to content

Guides

Open this in the pixkoz appGet the pixkoz app on Google PlayUse the pixkoz app

OpenGetOpen

Is it safe to upload my photo to AI apps?

By the pixkoz team · Updated · 5 min read

Mostly yes, if you know where the photo goes and set the right switches. An AI photo app has to receive your photo to edit it, so the real questions are who keeps it, for how long, what they may use it for, and who else can see the result. This guide answers them for Gemini, ChatGPT and pixkoz, as of October 2026.

We run pixkoz, so for pixkoz we can tell you exactly what happens; it is all in our privacy policy. For Gemini and ChatGPT we describe what their settings say today. Check their own privacy pages too, because they change.

Where your photo goes

  • When you copy a free prompt from pixkoz, your photo never touches pixkoz. You add it inside Gemini or ChatGPT, under that company's terms.
  • When you cast a look on pixkoz's website, your browser first shrinks the photo and removes hidden details such as its location (EXIF). We store it and send it, with the look's prompt, through OpenRouter to the AI model the look uses, Google's Gemini or OpenAI's GPT, which make the result and send it back.
  • The pixkoz Android app sends no photo to any AI provider. Its only upload is a profile picture, if you add one.

Gemini and ChatGPT: the settings to check

Both companies may use what you upload to improve their AI unless you turn that off. At the time of writing:

  • Gemini: the Keep Activity setting (called Gemini Apps Activity until 2025). When it is on, Google may use a sample of your chats and uploads to improve its services, with help from human reviewers, and chats a reviewer has seen are kept apart even if you delete your activity. Turn it off, or use a temporary chat, for photos you'd rather keep private. Google's Gemini Apps Privacy Hub has the details.
  • ChatGPT: in Settings, Data Controls, the switch Improve the model for everyone. When it is on, your chats, photos included, may be used to train OpenAI's models. A Temporary Chat stays out of your history and, OpenAI says, is not used for training, though it may be kept for up to 30 days for safety checks. See OpenAI's privacy policy.

Switching a setting off applies to what you upload from then on. To remove older chats, delete them in the app as well.

What pixkoz does with photos you cast

  • We do not use face recognition, and we create no face templates or other biometric identifiers.
  • We do not sell your photos or results, and we do not use them to train AI models of our own.
  • The photos you cast and their results are not shown to other people on pixkoz. They are kept until you delete your account, on our image storage at long, unguessable links, so anyone you give a link to can open that image.
  • Deleting your account deletes them. Deleting a single cast is not possible yet.
  • Every AI picture pixkoz shows is labelled AI-generated, and the files carry embedded information that marks them as AI-made, with nothing about you in it.
  • pixkoz is for adults aged 18 and over.

The full details, and your rights, are in our privacy policy.

Photos not to upload to any AI app

  • Other people's photos without their OK. Use photos of people who have agreed.
  • Photos of children. Many AI apps restrict them, and pixkoz is for adults only.
  • Photos with an ID card, a passport, a bank card, a boarding pass or your house number visible, even in the background.
  • Intimate or revealing photos. Even with training switched off, they are stored somewhere you don't control.
  • Screenshots that show other people's chats, numbers or email addresses.

A 30-second check before you upload

  1. Is it a photo of me, or of someone who said yes?
  2. Is anything private in the frame: a document, a screen, an address?
  3. Is the app's training setting the way I want it?
  4. Would I be fine if this photo became public? If not, don't upload it.
  5. Do I know how to delete it later?

Will people know the picture is AI-made?

They can. Google marks images made in Gemini with SynthID, an invisible watermark, and OpenAI adds Content Credentials (C2PA) to images from ChatGPT. On pixkoz, every AI picture is labelled AI-generated on screen, and the files carry embedded data that marks them as AI-made. Being open about it is good manners anyway: say a picture is AI-made whenever it could pass for a real photo.

Deepfakes and misuse

Never use an AI app to put someone in a picture they did not agree to, to make it look as if they did something they didn't, or to make anything sexual of a real person. It hurts people, much of it is against the law in India and in many other countries, and all of it breaks the rules of every major AI app, pixkoz included. If you see it on pixkoz, report it with Report in the look's menu.

Questions people ask

Does pixkoz see my photo when I use a free prompt?

No. You add your photo inside Gemini or ChatGPT; pixkoz only gives you the prompt text.

Can I delete a photo I cast on pixkoz?

Deleting your account deletes your photos and results. Deleting a single cast is not possible yet.

Does pixkoz use my face to train AI?

No. We do not use your photos or results to train AI models of our own, and we do not use face recognition.

So, is it safe to upload my photo to Gemini or ChatGPT?

For an ordinary photo of yourself, most people are comfortable once they have checked the training setting. Keep private and sensitive photos out of every AI app.

More guides

All guides