The strange thing about huge, insanely wealthy corporations is that they sometimes come across as tone deaf when it comes to communicating with their customers (i.e. the people that made them wealthy). We are not saying they don't care, it's either that they: don't run new policies through focus groups; or they implicitly trust lawyers whose 'law jargon' often flies over the heads of consumers (and sometimes the actual company); or they just don't use the 'flow chart of possible outcomes' that if used properly, could pick up some of the policies that are going to cause backlash.
We are at an interesting point in image making history at the moment, and many content creators (photographers + videographers) are feeling threatened by AI. So it comes as no surprise that when Adobe miscommunicates that customers images may (or may not) be used to train AI – there is going to be some backlash. After all – who wants to see snippets of your own work somehow appear in AI images and then not be rewarded for it?
For an example on how AI-generated images can work, the Canva website explains the way they come up with AI-generated images as below.
"To create AI-generated images, the machine learning model scans millions of images across the internet along with the text associated with them. The algorithms spot trends in the images and text and eventually begin to guess which image and text fit together. Once the model can predict what an image should look like from a given text, they can create entirely new images from scratch based on a new set of descriptive text users enter on the app."
So while Canva (primarily graphic design) doesn't tout themselves as a photography site, for Adobe it's different as they have been seen as a photo industry champion for decades – hence the blowback...
Last week Adobe announced that they are making changes to Terms of Use around how they use data to train generative AI. This caused an uproar and led to them having to reconfigure and clarify actually how their AI-generation system works.
The question begs - as a photographer, are you feeling threatened by AI yet? AI image: Tim Levy
After a blog post late last week, Adobe has now published another post, this time announcing plans to better communicate the changes.
“We recently rolled out a re-acceptance of our Terms of Use which has led to concerns about what these terms are and what they mean to our customers. This has caused us to reflect on the language we use in our Terms, and the opportunity we have to be clearer and address the concerns raised by the community,” Adobe says in a new blog post written by Scott Belsky and Dana Rao. Belsky is Adobe’s Chief Strategy Officer, and Rao is Executive Vice President, General Counsel, and Chief Trust Officer.
Much of the concern was around the suggestion that Adobe planned to use user data to train AI datasets. However Belsky and Rao are unequivocal that user data will never be used to train any generative AI tool.
“We’ve never trained generative AI on customer content, taken ownership of a customer’s work, or allowed access to customer content beyond legal requirements,” say Belsky and Rao.
"We will make it clear in the license grant section that any license granted to Adobe to operate its services will not supersede your ownership rights."
Adobe's Firefly AI tool is, according to the company, only trained on a dataset of licensed content with permission, such as Adobe Stock, and public domain content where copyright has expired.
“At Adobe, there is no ambiguity in our stance, our commitment to our customers, and innovating responsibly in this space,” write Belsky and Rao.
However, according to Adobe, customers should expect 'significant clarification' regarding content ownership, training generative AI models, usage licenses, and content moderation in the upcoming update in an update to its Terms of Use it describes as “the right thing to do.”
Revisions to the Terms of Use will focus on numerous areas, including how Adobe trains generative AI models, treats user content, and moderates content.
Adobe says that while user data is used to help improve some machine learning features, users can always opt-out.
"In a world where customers are anxious about how their data is used, and how generative AI models are trained, it is the responsibility of companies that host customer data and content to declare their policies not just publicly, but in their legally binding Terms of Use," write Belsky and Rao.
