Is Adobe training AI on your photos? It's complicated

A sharp-eyed developer working on the open-source drawing tool Krita recently spotted something fishy in Adobe Creative Cloud account settings. It turns out Adobe had automatically opted every user into a program called content analysis.

The policy explicitly states that Adobe may analyze your content using techniques like machine learning to develop and improve its products and services. Naturally, creative communities immediately panicked, assuming Adobe was quietly feeding their private artwork into AI models.

Adobe quickly stepped in to clarify. Speaking to PetaPixel, the company claimed this policy is not new at all and has actually been in place for a decade. But in an era where everyone is terrified of their creative work being scrapped by algorithms, that explanation did not really calm anyone down.

Training algorithms versus training generative models

So is Adobe training AI on your pictures? The short answer is yes, but with a major catch.

Adobe stated that it does not use any data stored in customers' Creative Cloud accounts to train its experimental generative AI features. That phrasing is notably precise. While it rules out training image generators, it does not rule out machine learning altogether.

In fact, Adobe's own documentation confirms that it aggregates user content with other files and uses the combined data to train its algorithms. According to Adobe, this powers utility features like automatic object recognition for pets in Lightroom, perspective correction in Photoshop, and general product analytics. In some cases, content might even be subject to human review.

While Adobe also has a separate, opt-in Photoshop Improvement Program, the default content analysis toggle was turned on for everyone without asking.

Why it matters

We think the core issue here is not the technology itself, but the sneakiness of opting users in by default.

Software telemetry is common, but analyzing user files to train machine learning models crosses a clear line for professional creators who deal with proprietary work, strict NDAs, and commercial assets. Claiming the policy is a decade old does not help Adobe's case; it only highlights how long users were kept in the dark.

Moreover, the careful legal phrasing leaves plenty of room for skepticism. Saying data is not used for experimental generative AI features feels like a loophole waiting to be exploited once those features leave the experimental stage.

If you use Creative Cloud, our advice is simple: log into your Adobe privacy settings right now and manually toggle content analysis off. For Adobe and other creative software vendors, this backlash shows that opacity around user data will no longer fly in the age of AI.