A recommendation preference, not a complete filter
Pinterest announced new GenAI-content preferences on October 16, 2025. Users could ask to see fewer eligible AI-generated or AI-modified image Pins in supported categories such as art, beauty, fashion and home decor. The announcement described Android and desktop availability, with iOS following gradually.
The controls were accessible through recommendation settings or a Pin's overflow menu, and users could revise their selections. A December update added food and drink to the supported categories. That later addition should not be treated as part of the original October launch scope.
Labels provided a separate transparency mechanism
In April 2025, Pinterest had already announced labels for images its systems detected as generated or modified with AI. The company described examining metadata and using classifiers to identify images even when obvious markers were absent. The labels appeared when a user opened an image Pin in closeup.
The same earlier announcement acknowledged that detection was imperfect and described an appeals route for people who believed an image had been mislabeled. It also previewed a see-fewer experiment. Those details distinguish two related functions: identifying suspected AI modification and changing how often similar content is recommended.
User control depends on the scope of detection
The October tuner gave users a more explicit way to express preferences, but neither announcement promised to remove every AI-created image. A category-based preference acts on eligible content that the platform can identify; it is not the same as verifying the origin of every image. The acknowledged possibility of mislabeling also matters for creators whose work may be affected. Together, the announcements describe a progression from labels and experiments to broader preferences, rather than a blanket ban on AI content or proof that the recommendation system always recognizes it correctly.