2024-05-07 06:30:45
False content has become more and more numerous since the “democratization” of generative artificial intelligence. To better inform Internet users, the social network will begin in May to affix “Made with AI” labels to videos, images and sounds generated by this technology and published on its platforms.
Some find them funny or don’t pay attention, others are shocked and worried. AI-generated videos abound on the internet. As any election approaches, we see such clips appear, which present themselves as parodies or schoolboy diversions. But the risk of misinformation is permanent.
All countries are affected. A fake AI-generated interview broadcast days before Slovakia’s September 2023 parliamentary elections attempted to present Progressive Slovakia leader Michal Šimečka as having discussed with a journalist how to rig the vote.
Fake audio content has falsely implicated a presidential candidate in plans to manipulate ballots in Nigeria’s February 2023 elections. In Bangladesh, doctored videos of opposition members in bikinis and in a swimming pool have were published before the national elections.
France is not spared from deepfakes. Two TikTok accounts broadcasting videos young girls, exposed navels and long blond hair, with the faces of Marine Le Pen and Marion Maréchal, were deleted.
Disinformation and fake news
Disinformation, AI-generated content and fake news pose a serious threat to democratic processes as half the world’s population participates in more than 80 national elections in 2024 – including the European elections in June.
To stem this scourge, the European Parliament reached an agreement in December 2023 on the proposed European Media Freedom Act (EMFA). It aims to harmonize the national laws of EU member states on editorial freedom, pluralism and media independence.
Meta, Facebook’s parent company, is at the forefront of this scourge. After years without real action, the group announced that it wanted to label the videos broadcast on its platforms to differentiate “real” videos from those generated by AI.
The labels will be applied either when users disclose the use of AI tools or when Meta detects “industry standard AI image indicators.” However, the company did not provide further details regarding its detection system.
Furthermore, it is not yet clear what impact “deepfakes” might have on elections and we might even see this type of content used in ways that have not yet been imagined.
Finally, it is also worth remembering that disinformation is not always a high-tech affair. There are other ways to attack or destabilize a democracy: rumors, conspiracy theories, electoral fraud, etc.
Even if Meta’s announcement is to be welcomed, no system can be perfect in detecting deepfakes. Errors, or even abuse, might be noted. In 2018, Facebook censored the bare breasts in “Liberty Leading the People” and “The Origin of the World” by Gustave Courbet…
1715296968
#Facebook #inform #Internet #users #deepfakes
