Product reviews, deals and the latest tech news

Do Adobe’s AI training algorithms use your photos? It’s challenging

A developer at Krita recently noticed that, in the Adobe Creative Cloud settings, the company had opted them (and everyone else) into a “content analysis” programme, wherein Adobe “may analyse your content using techniques such as machine learning (for example, for pattern recognition) to develop and improve our products and services.” Some have taken this to suggest that it is assimilating your photographs for its AI. Of course they do So, in a sense? But things aren’t quite that easy.

To begin, a great deal of currently available software has a “share information with the developer” option that transmits telemetry like how often the app was used, which features were used the most, why the programme failed, etc. In most cases, you’ll be given the choice to disable this feature during setup; however, Microsoft has drawn criticism for making telemetry the default setting in Windows 10 and making it hard to disable.

That’s disgusting, but it’s even worse to secretly implement a new form of sharing without alerting your people. Adobe stated to PetaPixel that their content analysis system “is not new and has been in place for a decade.” It’s remarkable if they’ve been utilising machine learning for this reason and have been vocal about it for a decade, and even more remarkable that no one has noticed. It’s hard to believe that would happen. It’s likely that this policy has been around in some form, although one that has changed subtly over time.

It may use machine learning to evaluate your material, but not for training purposes; the setting makes this explicit. Following the “learn more” link, you will get the following information:

For example, we may use machine learning-enabled features to help you organize and edit your images more quickly and accurately. With object recognition in Lightroom, we can auto-tag photos of your dog or cat. In Photoshop, machine learning can be used to automatically correct the perspective of an image for you.

Adobe would also be able to identify, using a machine learning study, what percentage of Photoshop users are primarily adjusting portraits as opposed to landscapes. That could help direct the development and prioritisation of new products.

The photographs and analyses may be used to train AI models as part of “improving our goods and services,” as you correctly point out.

While this is true, Adobe has said, “Adobe does not use any data kept on users’ Creative Cloud accounts to train its experimental Generative AI capabilities.” The language is straightforward enough, but it has an air of legal accuracy that makes you wonder if they’re trying to skirt around an issue.

Further inspection of the accompanying documentation reveals the following claim: “When we analyse your content for product improvement and development purposes, we first aggregate your content with other content, and then use the aggregated content to train our algorithms, and thus improve our products and services.”

Thus, it does make use of user-provided data in order to fine-tune its algorithmic processes. Maybe not so much its unproven Generative AI algorithms.

In reality, Adobe has a programme designed for that same purpose; the Adobe Photoshop Improvement Program, which is described in detail here and requires the user’s consent. However, it’s feasible that your photographs are being utilised in some way to feed data into a generative AI. There are additional cases when human review may be necessary, which is a different matter entirely.

If you care about your privacy, you should avoid Adobe and any other company that collects your data in this way. If you are a registered user, you have the option to do so on the privacy page.