Facebook and Instagram Reduce Visibility of AI Labels on Edited Content

Meta is enhancing how it labels content on Instagram, Facebook, and Threads that has been edited or manipulated using generative AI. In a recent blog post, Meta announced that its new “AI Info” tag will appear in a menu in the top-right corner of AI-edited images and videos, rather than directly beneath the user’s name. Users can access this menu to find information about the AI adjustments made to the content.

Previously, the “AI Info” tag was applied broadly to all AI-related content, regardless of the level of editing, from minor changes in tools like Photoshop to fully AI-generated creations. Meta states that these updates aim to more accurately reflect the extent of AI involvement in the content shared across its platforms.

The label was first introduced in July following criticism of Meta's prior “Made with AI” tag, which incorrectly labeled authentic photographs captured by creators and photographers. Meta clarified that it will continue to display the “AI Info” label for content generated by AI tools, indicating whether this labeling is due to industry-standard signals or self-disclosure by users. These changes are set to begin rolling out next week.

The term “industry-shared signals” refers to systems such as Adobe’s C2PA-supported Content Credentials, which can be applied to content created or edited with its Firefly generative AI tools. Other systems, like Google’s SynthID digital watermarks, are also used for content generated by its AI tools. However, Meta has not revealed which systems it monitors for compliance.

While these updates aim to improve transparency, completely removing tags from real images that have been edited could complicate efforts for users to avoid potential misinformation, particularly as generative AI editing tools become more sophisticated on newer devices.