Washington (AFP)
Facebook will ban "deepfakes", these videos manipulated using artificial intelligence to replace a face or change the words of a personality, announced Tuesday the social network in a blog note.
In this note, published after information to the same effect from the Washington Post, Facebook states that the modified videos will be deleted without it being obvious, likely to "mislead" Internet users about the words of a personality, and which are the product of an artificial intelligence which "mixes, replaces or superimposes" a content in a video by making it appear "authentic".
This decision, announced in this year of presidential election in the United States, "does not apply to parodic or satirical content" or to videos which have been edited in such a way as to omit words or to change their order, specifies the author of the note, Monika Bickert, vice president of Facebook.
Video like the one suffered by Nancy Pelosi, president of the US House of Representatives, where she appeared to be drunk due to slower reading speed, would escape these new guidelines, American media said. .
Facebook will also continue to remove audiovisual content that violates "community standards", and greatly reduce the audience of those labeled as false by the external fact-checkers with whom it works, including AFP. .
AFP participates in nearly 30 countries and 10 languages in "Third party fact-checking", a third-party verification program developed by Facebook since 2016. With this program, Facebook pays around sixty media around the world, generalists or specialized, for the use of their "fact-checks" on its platform and on Instagram.
© 2020 AFP