Meta's social networks to label photos made with artificial intelligence: how it will work

On Facebook, Instagram, and Threads, all images created or edited with the help of artificial intelligence will be marked with appropriate labels. The AI content identification will appear soon and will be extended to video and audio.

This was announced by Meta, the owner of these networks. The statement says that the markers will be applied "in the coming months after identifying standard industry indicators of artificial intelligence created".

Meta has not yet decided what the "Created by AI" labels will look like and where they will be placed: "We are just creating this feature now, and in the coming months, we will start applying the labels in all languages supported by the applications."

In addition to visible markers, experts promise invisible "watermarks" on photos and metadata that will be embedded in image files. C2PA, IPTC metadata and watermarks will be in line with the best practices of the Partnership for Artificial Intelligence PAI.

The company clarifies that the labeling process will continue "throughout the next year around the world." Meta emphasizes this as a number of important elections will be held in many countries.

The company also promises to work with Google, OpenAI, Microsoft, Adobe, Midjourney, and Shutterstock to add metadata to images created by their tools.

Meta adds that it is "not yet able to detect and tag all content created by other companies' AIs." Moreover, "there are ways that people can remove invisible markers," the experts warned.

Subscribe to OBOZREVATEL on Telegram and Viber to keep up with the latest news