For the first time, a ‘deepfake’ was used in a pro-Beijing propaganda video

TV presenters, who appear larger than life but are actually generated by artificial intelligence, have been used in propaganda videos for the first time, says a US research firm, claiming the actors defend their pro-Beijing positions .

The latter, who “work” for a fictitious television news called Wolf News, are actually products of artificial intelligence, which appeared on social networks, apparently to promote the interests of the Chinese Communist Party, indicates the US firm Graphika in a study . .

“Creating misleading political content”

This hyperfaking or “deepfake” makes it possible, in particular, to substitute one face for another in a video or to falsify the words of a character with disconcerting realism.

In one of these videos analyzed by Graphika, a presenter who calls himself Alex criticizes the inaction of the United States in the face of the gun violence that is ravaging this country. In another, a presenter stresses the importance of “cooperation” between the great powers of China and the United States.

Advances in artificial intelligence have fueled fears that this technology could also be used to spread fake news and “deepfakes”, these increasingly realistic digital manipulations, which present a challenge in the fight against disinformation.

A “danger to national security”

These statements by Graphika, which have not been commented on by Beijing, come a few weeks after the Chinese government announced new rules to regulate “deepfakes.”

Hyperfaking poses a “danger to national security and social stability” if left unchecked, China’s internet policeman estimated in December. Regulations now require companies offering deep-spoofing services in China to obtain the real identities of their users.

For Graphika, the two Wolf News presenters were undoubtedly created thanks to technology provided by London start-up Synthesia. This company, which did not immediately respond to AFP’s requests, sells software to create avatars “based on live-action videos,” according to the company’s website.