Recently, this social network announced that it will label "AI modified" (EDited with AI) in the lower left corner of images created or edited with artificial intelligence.
In a post announcing the new feature, Pinterest said it had been quietly testing the tool for several months, by analyzing the metadata of the photo to determine its origin. In addition, the company is also developing an AI-generated image recognition algorithm even when there are no signs in the metadata.
Pinterest admits that the system's accuracy will require time to verify when widely deployed. Content creators who think their photos are mislabeled have the right to complain.
Incorrect labeling has been a problem for Meta - the parent company of Facebook and Instagram, as many photographers reacted because their original photos were labeled with AI. Pinterest will definitely have to avoid this traffic jam if they want to keep the quality.
Pinterest is also testing a new feature that allows users to proactively filter out categories that are often overused by AI, such as beauty or art. With this step, the platform hopes to regain the trust of the community, who are increasingly frustrated by the wave of fake content.
