Meta finally has an answer to all those asking why it labels photos with minimal edits as “Made with AI.” The parent company of Facebook and Instagram, is making a change to how it identifies photos potentially edited with artificial intelligence (AI). It is replacing the previous “Made with AI” label with “AI info” tag.
This shift comes after users, especially the photographers, expressed confusion about the “Made with AI” label. In some cases, the label was mistakenly applied to photos with minimal edits or those not created with AI at all.
The new “AI info” tag aims to provide clearer context. Clicking on the tag will reportedly offer users more details about how AI might have been used in the photo. This could include information about automated adjustments like lighting correction or background enhancements.
“We’ve found that our labels based on these indicators weren’t always aligned with people’s expectations and didn’t always provide enough context,” Meta acknowledged in the blog post.
While the specific functionalities behind the “AI info” tag remain to be seen, the change seems to foster transparency and combating potential misinformation. By offering more details about potential AI manipulation, users can be better equipped to critically evaluate the content they see on Meta platforms.
Meta also acknowledged feedback from the Oversight Board that their current policy on manipulated media is very limited. Initially, the policy focused primarily on AI-altered videos that make a person appear to say something they didn’t actually say. This approach was established in 2020 when such AI-generated content was rare and the concern mainly revolved around video formats.
However, Meta now recognises that technology has evolved rapidly, leading to the creation of other types of manipulated media beyond videos, such as realistic AI-generated audio and photos. These newer forms of manipulated content also pose significant challenges in terms of misinformation and deception.
The Oversight Board also advised Meta to update its policy to encompass a broader range of manipulated media formats and to provide context about such content through labelling. Meta agrees with this perspective, indicating that they will expand their policy to address manipulation that shows a person doing something they didn’t actually do, not just saying something they didn’t say.
