Meta Modifies Its AI Labeling System Following Backlash
Professional photographers have criticized Meta over its AI labeling system, which tagged edited real-life photos as AI-generated content.
In an effort to curb AI misinformation on its platforms, Meta introduced a labeling system that would flag AI-generated content as “Made with AI.”
The Unforeseen Consequence
What the team did not expect was the AI labeling system flagging any edited picture, regardless of whether it was a real-world photo. The system, for example, flagged photos that were touched up using tools like Photoshop. This has garnered enough criticism from professional photographers that Meta decided to do something about it.
Former White House photographer, Pete Souza, pointed out how an image, taken 40 years prior, was labeled by the AI labeling system once cropped and flattened using Adobe software. Souza speculated that these simple edits might have triggered the label.
Meta spokesperson Kate McLaughlin addressed the situation, stating, “As we’ve said from the beginning, we’re consistently improving our AI products, and we are working closely with our industry partners on our approach to AI labeling.” The revised label aims to clarify that the content might only be modified by AI tools, rather than being entirely AI-generated.
All’s Well That Ends Well
After Meta expanded its AI content labeling policies, platforms such as Instagram, Facebook, and Threads began tagging retouched real-life photos as “Made with AI.”
Users will first encounter the updated label on mobile apps, with web updates to follow. According to McLaughlin, the new labeling is gradually rolling out across all Meta surfaces.
Despite the change, clicking the new “AI info” tag will still display the same detailed message as before, explaining why the label was applied. This message indicates that the image could either be fully AI-generated or edited with AI-enhanced tools, such as Adobe’s Generative Fill.
Despite this situation being resolved, there is the problem of how platforms interpret the metadata from editing tools like Photoshop. Originally, metadata tagging technologies like C2PA were designed to simplify distinguishing between AI-generated and real images. However, the implementation of this future-proofing measure has yet to achieve its intended clarity.
Inside Telecom provides you with an extensive list of content covering all aspects of the tech industry. Keep an eye on our Tech sections to stay informed and up-to-date with our daily articles.