How to Edit Your Photos in ChatGPT (and Keep Your Face)

Who can use ChatGPT for photo edits

OpenAI's help page says ChatGPT Images is available on all tiers and on web, iOS and Android. A version called Images with thinking is listed for paid plans such as Plus, Pro and Business. Usage limits can apply and they change from time to time, so check the current help page if you hit a cap.

OpenAI's terms of use say you must be at least 13 years old, or the minimum age in your country, to use its services, and if you are under 18 you need a parent or guardian's permission. The help page also notes that a parent or guardian with a linked teen account may be able to control whether the teen can create or edit images.

How to edit a photo in ChatGPT, step by step

These steps describe the general flow as of October 2026. The labels may differ slightly on your phone or browser.

1. Open ChatGPT and start a new chat. Older instructions in a long chat can affect the new image.
2. Attach your photo with the upload or plus button beside the message box. Pick a sharp, front-facing picture with good light.
3. Write your request. Name the change first, then what must stay the same. For example: "Change the background to a rooftop with fairy lights at night. Keep my face, skin tone, hairstyle and shirt exactly as they are."
4. Add the shape you need. OpenAI's help page says you can use the aspect ratio picker or include the ratio in your prompt, such as "vertical 4:5".
5. Send and wait. The help page notes that generation can take a few minutes for complex requests, and you can keep chatting meanwhile.
6. Look at the result closely: eyes, teeth, ears, hands, jewellery and the edges of your hair.
7. Ask for one change per message, such as "make the lights warmer" or "remove the plant behind me".
8. Save the image. On mobile the image controls include Save, Copy and Share, according to the help page.

OpenAI's help page says images you create are saved under Images in the sidebar, and to delete an image from there you delete the conversation where it was made.

Using the selection tool for small fixes

OpenAI describes two ways to edit: select part of the image with the selection tool and then describe the change, or skip the tool and describe the edit in the chat. The selection tool is handy when only one spot is wrong, like a sixth finger or a stray object.

1. Open the image you want to change from the chat or from Images.
2. Choose Select and brush over only the area you want changed. On mobile there is a slider for brush size.
3. Tap Next or move to the text box and say what should happen in that area, for example "replace with a normal hand holding the diya".
4. Use Undo or Redo if your selection spills over, and Save when you are happy.

OpenAI's help page warns that highlights are not always precise and edits may extend beyond the area you selected. So after a selected edit, check your face again, even if you did not brush over it. If you did not use the tool, name the area in words ("only the background on the right side").

How to edit a photo in ChatGPT without changing your face

When you upload a portrait, the model makes a new image based on it. That is why small details like the shape of your nose or your hairline can shift. You cannot fully control this, but clear wording helps the model understand your priority.

• Put the keep-list in every message. "Keep my face, age, skin tone and hairstyle the same" works better repeated than assumed.
• Ban beautifying. Write "no skin smoothing, no slimming, no makeup change" if you want to look like yourself.
• Change one thing at a time. Outfit first, then background, then light. Big combined changes give the model more room to redraw your face.
• Describe Indian details precisely. "Kanjeevaram silk saree in deep green with a gold zari border" beats "nice traditional dress". Our saree prompts and traditional wear prompts show the level of detail that helps.
• Name the camera feel. "Natural photo, 85mm portrait lens look, soft window light" keeps the result closer to a real photograph.

For a full list of tricks, read our guide on keeping your face the same in AI photos.

Content Credentials and SynthID on ChatGPT images

OpenAI's help page on provenance says images generated with ChatGPT, Codex or the OpenAI API include two signals: Content Credentials and SynthID watermarks. Content Credentials are metadata built on the C2PA open standard, recording details like which tool created a file and when. SynthID is an invisible watermark embedded in the image itself.

The same page explains that Content Credentials can be lost through everyday actions like taking a screenshot or converting a file, while the watermark is designed to survive light edits but may become undetectable after heavy changes. It also says these signals do not guarantee that content is accurate or unedited, only where it may have come from.

We explain these marks so you know what is inside your file. We do not give steps to remove them. If you share an AI-edited photo, labelling it as AI-edited is the honest choice, particularly for anything that looks like news or a real event.

Common ChatGPT editing problems and quick fixes

More fixes for hands, colour casts and plastic skin are in our common AI photo problems guide.