Meta is revising its labeling for AI-generated images after facing widespread complaints from photographers about the previous “Made with AI” labels.
The update, prompted by user feedback, will change the label to “AI info” across all Meta apps, aiming to clarify the use of AI in photos and address the confusion caused by the original label.
What Prompted the Change?
The “Made with AI” label, introduced in response to criticism from the Oversight Board regarding Meta’s “manipulated media” policy, was applied to images that had undergone even minor edits using AI-powered tools.
Photographers reported that these labels were incorrectly applied to real photos with basic edits. Tests by PetaPixel revealed that minor modifications using Adobe’s generative fill tool in Photoshop could trigger the AI label, even for small changes like editing a “tiny speck.”
Meta acknowledged that the earlier labels did not align with user expectations and often lacked sufficient context.
The company will retain the underlying technology for detecting AI use, relying on industry standards such as C2PA and IPTC metadata. This means that images edited with AI tools like Adobe’s Generative AI Fill may still be tagged, but the new “AI info” label aims to better convey that the image might have used AI in the editing process rather than being entirely AI-generated.
The “AI info” labels will not provide specific details about the AI tools used or the extent of AI editing. A Meta spokesperson confirmed that the contextual menu accompanying the label would remain the same, offering a generic description of generative AI and noting that Meta may add the notice when AI signals are detected by its systems.
Meta and other social networks face the challenge of setting guidelines that do not unfairly penalize photographers who use AI-powered tools in their editing workflows. Concurrently, companies like Adobe should inform photographers about the potential for their images to be tagged on other platforms when using certain AI tools.
Meta hopes the new label will better align with user expectations and provide clearer information as the industry works to improve the labeling process.
Featured Image courtesy of Meta