Deepfake, because the audios of politicians and celebrities are a dangerous game

Every so often they come back, or perhaps they have never completely disappeared. In recent weeks, between the early political elections and the release of a new Netflix TV series, social networks have returned to talk about deepfake. Wanna Brands and Giorgia Meloni, but also Richard Benson and Giuseppe Conte: FakeYou – Deep Fake Text to Speech is a site that gives the possibility to create audio clips with the voices of some political and famous personalities. The reasons behind the development of the technology by the creator of the portal, Brandon Thomas, are fun and technological research.

However, on 12 October the Privacy Guarantor announced that it had opened an investigation that will investigate the “potential risks that could arise from improper use of personal datawhich is precisely the voice”As stated in the press release. The opening of the investigation reached The Storyteller company – Fakeyou, of Atlanta, which will have to better clarify the existence of the portal to the Authority. Of central importance they are how the items are constructedthe type of personal data used and the purpose of the treatment both well-known personalities and normal users who request the online service; as well as “the location of the data centers that store personal data, both with reference to registered users from Italy and to well-known personalities, and the technical and organizational measures adopted to guarantee a level of security adequate to the risk“.

Bruce Willis sells the rights to his image for the deepfake

Retired from the stage due to an aphasia problem, he is the first actor to grant this type of rights

Fake audio and video for a fee

In mid-September it is the voice of Silvio Berlusconi appeared, last week that of Matteo Salvini and Richard Benson, but also the voice of Cicciogamer89, a well-known youtuber and gamer. The creators of the vocal clips are Italian and the result is quite likely. At the time of writing there are twenty requests for new audio with these voices, and whoever manages the portal is willing to pay up to $ 150 to expand your library of Italian-language characters.

For less than half, $ 70, it is possible to sample the voice of a person, which is not necessarily that of the user who requests it. Among the options the user can define the origin of the voice in categories such as “family”, “customer”, “my voice”, “someone else’s”. Once you get the result it is difficult to know how users will use the required audios. The portal imagines some scenarios such as “create new songs“,”moderate the public stream on the Twitch platform“, and so on.