Meta recently updated the AI-generated image tags in the Instagram app, changing the previous "Made with AI" tag to "AI info" and adding more detailed information. The move was in response to user complaints about the incorrect application of AI tags. Former White House photographer Pete Souza also shared cases he encountered, showing that even traditional film photos may be misjudged as AI-generated content after being processed by image editing software. This update aims to improve the accuracy of labels and better explain the AI recognition mechanism to users.
On Monday, Meta announced that the "Made with AI" label in the Instagram app will be updated to "AI info", which users can click to view more detailed information. This move is to address user feedback that some photos that were not generated by AI were incorrectly labeled as "Made with AI". This issue appears to be related to image metadata and the platform's interpretation mechanism, and applications such as Adobe Photoshop may trigger this tag. The new "AI info" label will be launched on the mobile side first, and will be launched on the web side later. After clicking on the tag, users can view a more detailed explanation of what situations trigger the AI tag, including content entirely generated by AI and images edited using tools that incorporate AI technology. Meta’s move aims to improve user experience and better standardize the labeling of AI content.
The problem appears to be the metadata tools like Adobe Photoshop apply to the image and how the platform interprets the image. Real photos posted to platforms like Instagram, Facebook and Threads are being labeled "Made with AI" after Meta expanded its policy on labeling AI content.
Moving forward, you'll likely see the new tab first on the mobile app before rolling out to the web version. Once the label is clicked, it will still display the same message as the old label, with a more detailed explanation of why it might be applied, and that it can cover images generated entirely by AI or edited using tools containing AI technology, such as Generate padding.
By improving labels and providing clearer explanations, Meta is trying to find the best solution between balancing AI content recognition and user experience. We look forward to seeing a more accurate AI content recognition mechanism in the future.