© Unsplash/Arnaud Jaegers
This year, no less than 40 national or international elections are taking place around the world. In this context, the risks linked to disinformation are more present than ever. This is particularly the case with deepfakes and we know to what extent it is now possible to imitate the appearance and voice of a person to make them say or do compromising things.
Worrying uses of AI
Eileen Culloty, researcher at the University of Dublin, has written a very interesting article on this subject in online media The Conversation. The scientist cites in particular two recent examples where deepfakes were used to try to influence an election. In Indonesia, a video resurrected former President Suharto. The latter encouraged voters to go to the polls and this content caused controversy because it was created by the political party he led.
In Slovakia, the malevolence has gone up a notch, since a fake audio generated by AI emerged a few days before the legislative elections of September 2023. We heard the leader of the Progressive Slovakia party, Michal Šimečka, who was discussing with a journalist on the best way to rig the vote. Fortunately in this case, this falsification was of fairly poor quality and fact checkers were able to quickly establish its inauthenticity.
We also recently spoke to you about Amandine Le Pen, Léna Maréchal-Le Pen and even Chloé Le Pen on TikTok. These young women, blonde with blue eyes, pose as a member of the Le Pen family. In reality, they are created from scratch via AI to dust off the image of far-right parties in France.
What to do in the face of this danger which poses our democracies are in danger ? Eileen Culloty believes that social networks sometimes tend to ” allow or even fuel disinformation”. But according to her, we should not believe that everything is played out online.
And the researcher asks a series of questions that are, to say the least, relevant:
Is there an independent media system capable of providing quality investigations in the public interest? Are there independent electoral administrators and bodies? Exist are there independent courts to rule if necessary?? Are politicians and political parties sufficiently attached to democratic values rather than personal interest? ;? This election year, we may well find the answer to these questions.
Distinguishing the false from the real, mission impossible ?
In any case, recent progress in AI is not likely to reassure us. Last February we mentioned the case of Sora, the creation of OpenAI capable of generating ultra-realistic videos on command.
Hany Farid, researcher at the University of California in Berkeley alerted thus: “As is the case with other techniques in generative artificial intelligence, there is no reason to think that converting text to video will not continue to improve rapidly, bringing us closer to in addition to a time when it will be difficult to distinguish the false from the real».
Faced with risk, OpenAI has decided to restrict access to Sora and requested a group of “experts in areas such as misinformation, hateful content and bias” to study its nuisance potential.
📍 To not miss any news from Presse-citron, follow us on Google News and WhatsApp.
[ ]