Meta has announced a new update to its policies for labeling content generated or modified with artificial intelligence (AI) on the Instagram, Facebook, and Threads platforms. Previously, the label “AI Info” was applied directly under the user’s name for all AI-related content. However, this approach drew criticism from creators and photographers who complained that real photos were incorrectly tagged.
To solve – so to speak – this problem, Meta has decided to change the “AI Info” label. From now on, the elabel will be visible inside a menu in the upper right corner of AI-edited images and videos. Users can click the menu to see if there is AI information available and read what may have been edited.
The new AI info label is under this drop-down menu that we will all open when scrolling
This change is intended to more accurately reflect the extent of AI use in content shared across platforms. Meta said it will continue to display the “AI Info” label for content it believes was generated by an AI tool, indicating whether the label was applied based on industry-shared signals or was self-declared by the user.
The “industry-shared signals” Meta refers to include systems like Adobe’s C2PA-backed Content Credentials metadata, which can be applied to any content created or modified using Firefly generative AI tools. There are other similar systems, like the SynthID digital watermarks used by Google for content generated by its AI tools.
The new information provided under the AI Info menu
However, completely removing labels on real images that have been manipulated could make it harder for users to avoid being fooled, especially as the AI editing tools available on new smartphones are becoming more and more convincing. It’s a system that continues to raise doubts, in short, because not only is it far from foolproof, but in this new formulation it even seems to be relegated to the same visibility as the fine print in those 200-page Terms and Conditions contracts that nobody reads.
What do you think? Do you find these labels useful or do you think their presence is ineffective? Let us know in the comments below.