Adobe AI Model Training and Content Analysis
Questions and concerns about how Adobe uses user content for artificial intelligence and product improvement have become increasingly common in higher education. This article attempts to clarify Adobe’s policies around AI model training, content analysis, and data privacy.
Contents
Adobe ID's
- Every Evergreen account has a federated Adobe ID associated with it that uses the Evergreen email address and password to sign in.
- Users can also create a personal Adobe ID with their Evergreen email address (or any other email address).
- So one email address can be associated with two separate adobe accounts which can cause confusion.
Content Analysis
The Adobe Content Analysis FAQ contains information on opting out of content analysis.
Local Content
- "For clarity, all references to “content” in this FAQ are meant to refer to content processed or stored on our servers not your local device (like a computer, phone or tablet)."
Content Analysis for Product Improvement
- With a personal Adobe ID users can go to account.adobe.com/privacy and opt out of content analysis.
- Federated Evergreen ID's are automatically opted out.
- '"You have been automatically opted out from content analysis for product improvement when using Creative Cloud and Document Cloud apps. This setting does not apply when you separately choose to participate in product improvement programs."
- If users voluntarily participate in programs like beta releases, Adobe Stock, etc their data could be used for content analysis.
Generative AI
Adobe states in the Content Analysis FAQ
- "We do not analyze your content to train generative AI models, unless you choose to submit content to the Adobe Stock marketplace."
- "Our Adobe Firefly generative AI models were trained on licensed content, like Adobe Stock, and public domain content where copyright has expired."
There is additional information in the Adobe Gen AI Approach article.
- "Adobe Firefly models are trained on a dataset of licensed content, such as Adobe Stock, and public domain content where copyright has expired. Adobe Stock content is covered under a separate license agreement, and Adobe compensates contributors for the use of that content."
- "We do not train and have never trained Adobe Firefly on user content."
Concerns
Some people have brought up issues with training on Adobe Stock content. Users upload and license content to Adobe Stock. They only get paid when customers use the purchase the content via Adobe Stock. Adobe Stock pre-dates the Adobe Firefly model training. Some users uploaded content to Adobe Stock before AI training existed and the Adobe Stock license agreement was later modified. Some users had content on stock images sites that were later bought by Adobe. There is a Wired.com article on some of the issues.
Adobe has been making bonus payments to Adobe Stock users for the Firefly AI training.
