Pinterest is battling the onslaught of artificial intelligence junk that is increasingly clogging the platform after complaints from users. From now on, you’ll be able to see when pins of images appearing in your feed have been created or modified by AI. When users click on a close-up pin, the words “modified by artificial intelligence” appear in the lower left corner.
In a blog post published this week, Pinterest says that it has been testing a new feature for several months now that involves analyzing an image’s metadata to determine its source. The company is also developing classifiers that can automatically detect signs of generative AI even in the absence of metadata markers. How successful this will be will become clear when the new features are rolled out globally, and authors who suspect that their content has been mislabeled will be able to appeal.
Such mislabeling has been an ongoing issue for Meta, which was forced to adjust the wording of the AI labels it applied to uploaded photos on Facebook and Instagram after photographers complained that they were being added to images that were not created by AI. Apparently, even minimal use of the generative fill tool in Photoshop was enough to cause the label to appear. This is something that Pinterest will likely want to pay attention to as they hone this new algorithm.
In addition to the new labels, Pinterest says it is experimenting with a feature that will allow users to filter out categories where artificial intelligence modification or generation is particularly prevalent, such as beauty and art. By giving users more control over what appears in their feed, Pinterest obviously hopes to win back a growing number of disgruntled members of its community.
