Meta will start labeling AI-generated content across Facebook, Instagram and Threads

Meta has unveiled plans to collaborate with industry partners to establish universal technical standards for identifying AI-generated content, including video and audio. The company is already using “Imagined with AI” labels for photorealistic images generated by their Meta AI feature. However, the aim is to extend this labeling to content created with tools from other companies such as Google Bard, DALL-E, Midjourney, and more.

In the near future, Meta intends to label images posted on Facebook, Instagram, and Threads when they detect industry standard indicators of AI generation. To achieve this, Meta is developing advanced tools capable of identifying invisible markers on a large scale. These markers are the “AI generated” information in the C2PA and IPTC technical standards. This will enable Meta to label images from Google, OpenAI, Microsoft, Adobe, Midjourney, and Shutterstock as these companies integrate metadata into images produced by their tools.

The technology for labeling AI-generated content will depend on invisible watermarks and metadata embedded in files, according to Meta. Alongside this, Meta is introducing a feature that allows users to disclose when they share AI-generated video or audio, enabling them to add a label. This disclosure and labeling tool will be mandatory for users posting organic content featuring photorealistic video or realistic-sounding audio that has been digitally created or altered. Non-compliance with this requirement could result in penalties from Meta.