your content “will never be used to train any generative AI tool”

Last week Adobe, the company behind Photoshop, Premiere and Lightroom (among other programs), found itself involved in a huge controversy. The creators discovered that the most recent terms and conditions of use included a clause that allowed the company to “access all of your content through automated and manual methods,” including sensitive content.

The problem is that it was mandatory to accept those terms to use Photoshop tools, something the community did not seem willing to go through. Now the company has clarified the matter and has promised to leave everything explained in the future terms. But let’s start at the beginning to have all the context.

The new terms. On February 17, Adobe launched its new terms and conditions of use, a policy that users must accept to continue using the company’s tools. However, it was not until last week that what was stated in sections 2.2 and 4.1 was discovered. These sections state the following:

“Section 2.2. Our access to your content: We may access, view or listen to your content […] through automated and manual methods, but only to a limited extent and as permitted by law. For example, in order to provide the Services and Software, we may need to access, view or listen to your content to (a) respond to feedback or support requests; (b) detect, prevent or otherwise address fraud, security, legal or technical issues; and (c) enforce the conditions […]. Our automated systems can analyze your content and Creative Cloud client feeds […] using techniques such as machine learning in order to improve our services and software and the user experience.”

Section 4.1, for its part, defines what Adobe understands by content:

“Section 4.1. Content: “Content” means any text, information, communication or material, such as audio files, video files, electronic documents or images, that you upload, import, incorporate for use or create using the services and the software. We reserve the right (but not the obligation) to remove content or restrict access to the content, services and software if any of your content is found to violate the terms. We do not review all content uploaded to the services and. the software, but we may use available technologies, vendors or processes, including manual review, to detect certain types of illegal content (for example, child sexual abuse material) or other abusive content or behavior (for example, patterns of activity that indicate spam or phishing, or keywords indicating that adult content has been posted outside of the adult wall).

Problem? That Adobe explained this in an unclear way. Photoshop showed a pop-up to users notifying them of the changes and saying, simply, that they had been made to “clarify that we can access your content through both manual and automated methods, such as for content review.” This set off alarm bells for users, especially those professionals who manage and create confidential content.

Adobe defended itself. Faced with the controversy, Jérémie Noguer, product director of Substance 3D, explained on his Twitter/X profile that Adobe “is not going to access or read any project in any way” and that “it would make no sense to do so” since “all serious companies in the industry would abandon us immediately if that were the case.

The company, for its part, released a statement saying that it would not use user-generated content to train Firefly Gen AI, its generative AI model, and that it would never “take ownership of customer work.” This was a public commitment, but soon it will be something contractual and explicit in the terms and conditions of use.

Your content is yours, by contract. Yesterday, Adobe published a statement on its website claiming that “there is no ambiguity in its commitment” and that “we have never trained generative AI with client content, nor have we appropriated a client’s work, nor have we allowed the access to customer content beyond legal requirements.” Therefore, to make everything clear, they are going to update the terms and conditions this week, clarifying certain controversial aspects. The first two and most important are the following:

You are the owner of your content. Your content is yours and will never be used to train any generative AI tools. We will make clear in the licensing section that any license granted to Adobe to exploit its services will not supersede your proprietary rights.

We do not train generative AI on customer content. We are adding this statement to our terms of use to assure people that it is Adobe’s legal obligation. Adobe Firefly only trains on a dataset of licensed and permitted content, such as Adobe Stock, and public domain content whose copyright has expired.

In short. Adobe will not use user or user-generated content to train its AI, nor does it acquire ownership rights to it. This is, however, an issue that is far from over. The use of personal and user-created content to train AI tools and algorithms is undoubtedly going to be one of the most important debates in the coming months.

Image | Adobe edited by Xataka

In Xataka | You write a text, AI creates a new song: Adobe’s new project wants to conquer generative music