Skip to main content
Participant
November 18, 2025
Question

Photoshop + Nanobanana - Confidentiality and Data Usage Policy - NDA Work

  • November 18, 2025
  • 1 reply
  • 55 views

Hello,

I am seeking clarification on the data privacy and usage policies when using Photoshop's Generative Fill feature, when used with Nanobanana, for professional client work.

My primary concern is regarding highly confidential assets covered by a Non-Disclosure Agreement (NDA). If I apply Generative Fill on an existing, proprietary client design that has not been publicly released, I need to ensure the following:

1. Data Training: Will my original client image, the accompanying text prompt, or the generated output ever be used by Adobe or Google to train any AI models?
2. Third-Party Access: How does Adobe  ensure that this confidential data is not retained or accessed by third parties, including Google or other service providers, for training or analysis purposes outside of the immediate generation process?
3. Data Retention: What is the precise data retention policy for the input image and the generated content? Is the data immediately purged from Adobe's servers after the result is delivered to the user?
4. Recommended Settings: What specific in-app settings (e.g., in Creative Cloud or Photoshop Preferences) should I disable or verify to guarantee the maximum level of confidentiality and prevent any potential sharing or retention of my sensitive client material?

I understand Adobe states that Firefly is trained on licensed and public domain content, but I need concrete confirmation for these specific confidential use cases.

Thank you for providing reliable sources or official policy links to address these concerns.

1 reply

Community Manager
November 18, 2025

Hi @N385499802kn4,

Thank you for reaching out and for your questions regarding data privacy and usage policies for the Nano Banana feature. I have checked and can confirm that you have an active Teams license associated with the Adobe ID used to post this query.

Please be assured that Adobe does not train Firefly or any Generative AI models on customer content. This includes your original images, text prompts, and generated outputs. Firefly models are trained exclusively on licensed and public domain content, not on private, confidential, or NDA-protected assets. You may refer to the following documentation for more details:
https://adobe.ly/4i4FET3

Regarding third-party access: Generative Fill requires an active internet connection, which means your content is processed on Adobe’s secure servers. Adobe maintains strict security controls to ensure that your data is not shared with third parties such as Google for model training or analysis.

Regarding data retention: Firefly processing occurs in the cloud, and partner-model workflows may temporarily store selected regions of your content during processing. This is transient and required only for the operation of the feature. More details are available here:
https://adobe.ly/4rorJeT

Please note that the Generative AI User Guidelines indicate that prompts and results may be subject to short-term review to prevent abuse, which may involve temporary retention for moderation and safety purposes. You can review this here:
https://adobe.ly/4i4FET3

Regarding recommended settings: Since you are using a business profile or account (such as a team, company, or school account), you are automatically opted out of content analysis for product improvement. Additional information is available here:
https://adobe.ly/4o5Zy1c

If your organization upgrades to an enterprise contract, your administrator can further restrict the use of non-Adobe generative AI models. Reference details can be found here:
https://adobe.ly/4pgInLq

I hope this clarifies your concerns. If you have any further questions, I recommend contacting your administrator and requesting that they reach out to your Adobe account manager for additional assistance.

Regards,
^AN