Has been updated An update to what are essentially outdated terms of service has landed the Photoshop giant in a new hot potato over giving itself the right to review user content stored on its cloud.
The issue started online yesterday when video game concept artist Sam Santara pointed out a February update in an X post, wondering what it meant and asking if he was reading the sentence correctly.
“I cannot use Photoshop unless you grant me full access to everything I create in Photoshop, including NDA work,” Santara wrote, tagging Adobe for an explanation. A company representative responded, but first, the text in question.
Are all your artworks ours?
If we look at Adobe's General Terms of Use, we can see that they were last updated on February 17, 2024, with changes to two sections (2.2 and 4.1) that explain how Adobe accesses user content and how it defines that content, respectively. These are the paragraphs that have been criticized online by Santara and others unhappy with the update.
The updated terms state that Adobe may access content through automated and manual methods, but “only in limited ways and only as permitted by law.” This may include automatically scanning user content “using techniques such as machine learning to improve our services, software, and user experience,” Adobe noted.
This passage may be a sticking point for protesters, some of whom have threatened to switch to competitors or open source products. People are outraged that Adobe seems to be giving itself overly broad rights to use people's work however they like, but the expression is that Adobe is giving itself these boilerplate rights so that its products can handle people's material.
What content does Adobe scan? Apparently if it's uploaded into Adobe's systems it will be scanned.
“Content” means any text, information, communications or materials, including audio files, video files, electronic documents and images that you upload, import, embed or create. [Adobe] The makers of InDesign mention “Services and Software” in section 4.1. And, of course, this includes Document Cloud services and software.
On a page linked from the terms of service that covers Adobe's content analysis in more detail, the company says it analyzes content primarily to improve its service, but Adobe's terms of service also state that “we may use available technologies, vendors, or processes, including manual review, to screen for certain types of unlawful content,” such as CSAM or “patterns of activity indicative of spam or phishing.”
The company said content stored locally on the PC where the Adobe software is installed will not be scanned.
So Adobe can still scan your content, but anyone upset about the changes to their terms of service made in February should be aware that it was mostly a minor change that added clarifying language for the first time since 2022. That means Adobe has been able to use and scan your content in the exact same way for some time now, years in fact.
Additionally, there are limited opt-out options available.
Customers with personal Creative Cloud and Document Cloud accounts may disable content analysis, at least for product improvement and development purposes, using the steps outlined on the Content Analysis FAQ page linked above; business-level accounts have such functionality controlled by the terms of their contract with Adobe.
Scott Belsky, Adobe's chief strategy officer and EVP of design and emerging products, responded to Santara's concerns about X by reiterating his belief that Adobe won't train generative AI models on customers' content unless the customer provides it in cash.
But Belsky's words were not all reassuring, as he made clear that there are limitations to opting out.
“There will likely be situations where a company's terms of use allow some level of access (e.g. indexing documents to make them easier to find, updating components used from a CC Library throughout a document, etc.),” Belsky said.
Adobe warned late last year that it could face hefty fines for its subscription cancellation practices, but did not respond to questions about the matter.
June 7th Update
Adobe did not respond to questions, but RegistryThe company published a blog post yesterday addressing the issue, reiterating that it doesn't use customer content to train its AI.
“The focus of this update was to more clearly communicate the improvements we're making to our moderation processes,” Adobe said in a statement. “In light of the explosion of generative AI and our commitment to responsible innovation, we've added additional human moderation to our content submission review process.”
Adobe said it has a limited license to access users' content “solely for the purposes of operating or improving our services and software, enforcing our terms and complying with the law.”
