Skip to content

Use AI with care

Before you upload client photos to an AI tool

Find out where client photos go, how long they are kept and what your client allows before you use an AI editor.

The quick answer

Before you upload a client photo, check the rules for your tool and account. “We do not train on your images” answers only one question about your data.

Also check who receives the photo, how long it is kept and whether your client allows this.

In this article
  1. Ask four separate questions
  2. What “not used for training” tells you
  3. Check storage and review as well
  4. Use this check before the first upload
  5. A simple question to send your client
  6. Keep the same care after editing
  7. Sources

Keep these in mind

  • Check the tool, model and account you will actually use.
  • Check training, storage and sharing separately.
  • Get clear permission for private client work before uploading it.

Ask four separate questions

A photo may show a person, an unreleased product or a private location. Decide what needs protection before choosing an editing tool.

Start with these questions. A simple yes or no to “Is it private?” is rarely enough.

  • Training: Can the service learn from my photo, written request or result to improve its AI?
  • Processing: Which companies receive my photo to make the edit?
  • Storage: How long are the image, prompt and result kept? What does deletion remove?
  • Sharing: Is the result private by default, or can it appear in a gallery or shared link?

What “not used for training” tells you

It describes one use of your data. It does not, by itself, tell you that the image stays on your computer or is deleted as soon as the edit finishes.

Adobe says it does not use customer content to train its generative AI. Separate rules apply to work you submit to Adobe Stock. Adobe also says that a request to a partner model can send your instructions and needed files to that partner.

Both can be true: a service can receive your photo without using it for training. These are Adobe’s rules. Check other apps, plugins and accounts separately.

Read more: Adobe: Content Analysis FAQ · Adobe: Sub-Processors

Check storage and review as well

Find out how long files are kept and how to ask for deletion. Check whether backup copies stay longer. Ask the service if its rules do not explain this.

Check whether a person may see your image to help with a problem or check for misuse. Adobe’s AI rules, for example, allow software and people to review inputs and results for misuse and content checks.

“Not public,” “not used for training” and “never reviewed” are different promises.

Read more: Adobe: Generative AI User Guidelines

Use this check before the first upload

If the client has told you to keep the work private, follow their approved process. An editing job does not automatically give you permission to send photos to an outside AI service.

  1. Identify what the file contains. Include faces, private details and unreleased designs.
  2. Check the client’s instructions and any limits on outside services.
  3. Write down the exact app, model, account and plan you want to use.
  4. Read the current rules for who gets the files, how they use them and how long they keep them. Save the links and date.
  5. Ask the client or their data contact about anything unclear before you upload.

A simple question to send your client

You can keep the conversation clear and specific:

“I would like to use [tool and model] to change [part of the photo]. This would send [the file or selected content] to [service]. Are you happy for me to use it for this job, and are there any limits I need to follow?”

Fill in the brackets only with facts you have checked. If you cannot say what leaves your computer, resolve that first.

Keep the same care after editing

Check the result before sharing it. Remove private details that should not be in the final image. Share it only with the intended people. Keep or delete files as agreed.

Repeat the check if you switch tools, models or account types. Similar names do not mean the same privacy terms.

Sources & further reading

Product features can change. Check the linked documentation for the version and service you use.