Adobe is having a rough week. On Thursday, I reported that Photoshop users received a pop-up requiring them to agree to new terms that appeared to give Adobe access to their work. In response to the resultant outcry from furious creators, Adobe issued a response, clarifying their new terms of service document was largely the same as previous versions, with a few clarifying factors added to the update.
This only added fuel to the fire. Adobe wasn’t out of the blue demanding access to creators’ work; rather, they seemed to be saying, they already had that access. Adobe’s press release attempted to assuage concerns, saying that the company would only access cloud-based user data for three specific purposes: Features that required access to content (like generating thumbnails); cloud-based features, like Photoshop Neural Filters; and to look for illegal or otherwise abusive content.
The company claimed it would not access any data stored locally, and would not train any Firefly Gen AI models on user content. However, a deep dive into the terms of service reveals that Adobe takes cloud-based user content, aggregates it with other user content, and uses that to train its “algorithms.”
It all turned into a big mess (and a hit to Adobe’s stock price), which is likely why the company issued a second statement on Monday, while everyone was distracted by Apple’s WWDC announcements. Adobe says it is working on a new terms of service, including clearer language, that it will roll out to users by June 18. Importantly, the statement offers the following clarifications:
-
Adobe does not claim ownership of your content, and does not use your content to train generative AI.
-
You can opt-out of the “product improvement program,” which scrapes “usage data and content characteristics” for features like masking and background removal.
-
Adobe will explain the licenses they require you to agree to when using their products in “plain English.”
-
Adobe does not scan content stored specifically on your machine “in any way.” However, they scan everything uploaded to their servers to make sure they aren’t storing child sexual abuse material (CSAM).
None of this is really news if you’ve been following along. Adobe really wants you to know it doesn’t access the content stored locally on your computer, nor do they train their generative AI models using your content. However, they will train other AI models with your data—just not AI models responsible for creating anything. Great.
It’s good you can opt-out of that AI training if you wish, but it doesn’t change the fact that Adobe has demanded quite a lot of access to your cloud-based content. I’m repeating the same advice I gave in my last piece on this issue: If you need to use Adobe products and you don’t want the company accessing your work in any way, keep all your data local. That means storing all your Photoshop data on your computer or an external hard drive, rather than in the cloud. It’s less convenient, but much more private.
