The start-ups that combat the deepfake phenomenon

Listen to the audio version of the article

In a time marked by conflicts of all kinds, seeing an Eiffel Tower in the distance surrounded by tractors and bales of hay can generate disorientation. So if even today an image is worth a thousand words, the one created by Vincent Smadja sparked a myriad of posts, stories, videos and comments. A social tam-tam that also alarmed the local Parisian authorities. On the other hand, the risk was evident: in the days marked by farmers’ protests, those hotbeds of dissent across Europe added fuel to the fire of an already hot front. Question of public order and common sense.

Let’s go back to Vincent Smadja, the French AI artist known online by the nickname If Only and who became a celebrity by creating surreal scenarios thanks to artificial intelligence. All false, or almost. Because in his creations the monuments are real, but what is around them is not. So decoding the image becomes a difficult task. His Instagram page has over 90 thousand followers and every creation goes viral. «AI is not a simple innovation, but a revolution. It will bring new practices and even a new way of thinking and seeing reality especially among the younger generations,” he told the French press. That fake photo published and removed (but too late to avoid the multiplier effect) was also relaunched by the Instagram account with 12 million followers Thetrillionairelife and by the X Gen.Reality profile, close to the American Republican positions. They call it the “butterfly effect”. That flapping of wings in Brazil can cause a tornado in Texas, as the American mathematician and meteorologist Edward Norton Lorenz wrote way back in 1972.

World Economic Forum misinformation alert

But in the years marked by endless streams that tornado runs online, increasing in power. This is what the World Economic Forum also reported in the 2024 global risks report: misinformation is the number one danger. Followed by extreme weather conditions and critical changes in systems. The report is based on the opinions of more than 1,400 experts, politicians and international leaders. «Instability is dictated by polarizing narratives. The link between falsified information and social unrest is central,” commented Saadia Zahidi, CEO of the World Economic Forum. Nothing is as it seems in “deepfake”, a word coined in the very recent year 2017 as the definition of a human image synthesis technique based on artificial intelligence, used to combine and overlay existing multimedia content with original videos or images.

Fake news old and new

Thus hoaxes and scams multiply. Nothing new under the sun. More than three centuries ago (and long before social message boards) British coffee shops fueled fake news. In the Seventeenth Century, panic affected royal circles because these meeting places had become places of political dissent. So in 1672 Charles II issued a proclamation to contain the spread of false news that fueled gossip about the crown. «In the past only governments and the powerful could manipulate public opinion by presenting lies as truth. Today anyone with access to the Internet can do it because the very notion of truth has become fragmented,” said Indian writer Kenan Malik in the Guardian. «But fake photos are the least of the problems. When I teach my students the perception part of reality, I argue that it is a socially negotiated object. All this means that there is a narrative that feeds itself based on what you want to see. Today technology simply creates something plausible to fit within our window of attention. We don’t need something perfect, but something that is perfect enough to make us fall into error within a few seconds. The real problem is videos because they deceive us and we are not used to distinguishing the true from the false”, says Matteo Flora, professor at the London business school European School of Economics with the chair of security of AI and super intelligences.

Flora: «There is awareness: people doubt everything»

But the glass can also be half full. «Everything that is happening is spreading to end consumers. If previously these creations were the prerogative of a few government entities, now any user can create their own world. But there is greater awareness because people are starting to doubt everything, managing to intercept any artifacts, which some would not have happened before. So there is an evolution in doing debunking that involves more users. On the other hand, the immune system of social information adapts to any context: we have lost the illusion that images and videos are true”, specifies Flora. Meanwhile, start-ups are multiplying in the field to decode reality. Israeli Clarity has developed software to counter deepfakes and has just closed a $16 million round. «Deepfakes are a real threat. I am concerned by what is happening in the world with more than four billion people called to vote,” said CEO Michael Matias. But Italy also plays its game on an increasingly digital and global chessboard. Because if reality is increasingly “combed” – as Carlo Freccero claimed – it is up to innovators to try to decode it.