Adobe adds editing controls to ChatGPT: select an area, then adjust it yourself

On September 29, 2026, Adobe introduced interactive editing controls for its ChatGPT plugin. Select an object or area of an image, describe the change, then refine hue, saturation, contrast and other adjustments with the controls that appear. Editing happens in a side panel, and changes become context for the continuing conversation.

The update also includes direct text editing and highlighting in PDFs. Here we focus on an easier result to inspect: separating where a change applies from how far it goes. Selection identifies the object, language describes the goal, and manual controls set the degree.

The new interactions began rolling out globally on ChatGPT web and app on the announcement date. Having the Adobe plugin does not establish that every new control is visible for your account. If the relevant side panel is absent, check what is actually available instead of searching for menus from a different Adobe app.

When “a little more vivid” is hard to explain

Imagine using your own photo of a potted plant for a community swap event. The overall lighting is fine; you only want the pot’s color to stand out. Repeatedly asking for “more vivid” and then “too vivid” makes the degree hard to communicate. It can also obscure whether the leaves and background changed along with it.

Reduce the task to something visible: select the pot, request a saturation adjustment, then move the control while comparing the result. The object, the adjustment and your decision to accept it each have a place to be checked.

This can help people who know what looks wrong but lack photo-editing vocabulary. It does not judge truthfulness for you or guarantee that everything outside the selection stays identical. Compare factual details such as the location, people and item condition with the original. If the photo represents an item being offered for exchange, do not erase real wear to misrepresent its condition.

Manual adjustment lets you stop at a result you accept. For a simple crop or rotation, a familiar photo app may be faster. Complex layers, fine masks or professional print delivery still call for tools appropriate to those requirements.

Adobe’s setup guide explains how to add the plugin and invoke it in a chat. Start with one photo you have permission to process and that contains no sensitive information:

1. Add and invoke Adobe. Add Adobe from Plugins in ChatGPT. Start a new conversation and type @Adobe, selecting it from the menu, or choose the installed Adobe plugin through +. If your account lacks this entry, follow the interface actually available to you.
2. Check account requirements and the goal. Adobe says guests can use some tools; generative capabilities, Creative Cloud files and saving across sessions require an Adobe account. Adding the plugin does not mean every feature and unlimited usage are included. Keep the original and upload a copy.
3. Change one object. Open the image side panel, select the pot or brush over it, then request: “Use Adobe to adjust only the selected pot’s saturation. Let me inspect the controls first; do not change the item’s shape.” This is a trial requirement, not a guaranteed system outcome.
4. Adjust and inspect the boundaries. Make small changes with the controls actually shown. Compare the pot’s edges, leaves, background and real condition with the original. If selection or controls are unsuitable, fix that step before adding longer prompts.
5. Inspect the final file. Once satisfied, obtain the result through the saving or export action available in the interface. Reopen the file and check clarity and edit scope. A satisfactory chat preview does not mean a file is saved where you need it. Keep the original for future work.

One change to one photo is enough to test whether describing, adjusting and checking a file suits your task. The first attempt need not also replace the background, add copy, generate a poster and publish it to social media.