Meta identifies manipulated content posted on its social networks

In the run-up to the U.S. presidential elections, and as a video montage featuring President Joe Biden sparked debate across the Atlantic, the Supervisory Board set up by Meta enjoined the group to enable the identification of manipulated content.

On Tuesday February 6, Meta announced that it would soon identify any image generated by generative artificial intelligence, such as those from MidJourney, Open AI, Shutterstock, Microsoft, Adobe and Google. "In the coming months, we will label images that users post on Facebook, Instagram and Threads when we can detect industry-standard indicators that they are generated by AI," said Meta's head of international affairs.

The announcement comes in response to growing concern that manipulated content (whether AI-generated or not) could be used for disinformation purposes to influence opinion in the run-up to major national elections.

The rules and conditions of use of social networks will therefore be amended to ensure that images and sounds that have been manipulated by AI or other means are clearly identifiable by users, particularly when they pose a risk of harm to said users.

Meta had already introduced such identification for all content created using the Meta AI tool, as early as last December's launch of celui-ci.