What deepfakes are and why we should start fearing them

Deepfakes are a new application of artificial intelligence and machine learning capable of blurring the boundaries between true and false.

• The new frontier of hoaxes, of fake news, are deepfakes
• Artificial intelligence risks giving rise to urban legends that are difficult to undermine
• Some solutions to defend against this information drift

“What is the perception of reality?Is it the ability to capture, process and make sense of what our senses receive?If you can hear, see, touch or smell something, does that make it real?”This was asked in a video posted on Youtube last year Morgan Freeman, one of Hollywood's longest-serving and most famous stars.Except that, at the end of the clip, Morgan Freeman himself declared that he was not Morgan Freeman:what our senses as spectators had just received was a deepfake, that is, a new application of artificial intelligence and machine learning capable of blurring the boundaries between true and false.

What is meant by deepfake?

The term quickly colonized the media and much of the new common imagination, but what what deepfakes really are?To begin with, it should be noted that, although the world began talking about it in the years surrounding the Covid-19 pandemic, deepfakes had already existed and been progressing for almost a decade:in 2014 Ian Goodfellow, former head of Apple's artificial intelligence division, published a scientific paper which introduces the concept of "generative adversarial network" (or Gan, acronym for Generative adversarial network), a particular application of machine learning in which two neural networks "challenge" each other in a zero-sum game that allows them to learn to generate new imagesi starting from an existing data set.And in 2017, researchers from the multinational graphics card company Nvidia strengthen and make the results of the Gan networks more credible.

This is the birth of the deepfake:sites like thispersondoesnotexist.com they make available to the general public cheap falsification tools that are unprecedented in the history of humanity, and for the first time the world realizes that our relationship with reality is hanging by a thread.“Deep” because the neural networks of these generation tools are made up of different layers:is the deep learning, that form of artificial intelligence which allows us to replicate the features and voices of real people;“fake” because, well, there is no need for explanations.

In November 2017, an anonymous user posted an algorithm on Reddit capable of producing fake but realistic videos:the code ends up on the popular code sharing site GitHub, where it is made available to anyone for free.And since then deepfake makes its way into the first major industry in which he will come to prominence:porn.Applications like FakeApp allow anyone to insert anyone else's face into compromising videos that at first glance seem realistic, generating an infinite number of risks and a lot of misogyny:Scarlett Johansson, despite herself one of the most frequent protagonists of fake pornographic videos spread across the four corners of the internet, will say that fighting them is "a lost cause", given that "the internet is a dark black hole that feeds on itself".

What are the most famous deepfakes

Since its first steps, "deep fake" has come a long way, and some of its incarnations have led public opinion to question the merits and risks that this new technology brings with it:in addition to the aforementioned deepfake video of Morgan Freeman, the work of the Dutch Bob de Jong and Boet Schouwink, at least the fake surrender of Ukrainian President Volodymr Zelensky, circulated by Russia last year, and the deepfake of the actor Tom Cruise which became all the rage on Tiktok, created by special effects expert Chris Ume and Miles Fisher from hundreds of thousands of photos and videos of the Hollywood star.

The deepfake has also made its appearance on television, further widening the audience exposed to its spells:overseas, the satirical program Saturday Night Live aired a deepfake of Hillary Clinton in tears completely indistinguishable from a genuine video, while in Italy Striscia la Notizia did the same thing with an algorithmically generated video of former prime minister Matteo Renzi.In 2018, the online newspaper Buzzfeed had instead circulated a fake-but-real clip of Barack Obama, who revealed only at the end that he was played by the actor Jordan Peele.

The risks of deepfakes

With great power comes great responsibility, said a well-known comic book superhero, and the same applies to the unknowns that loom on the horizon of deepfakes.What happens when thousands, and potentially even millions, of people believe a fake video of a political leader announcing nuclear war?And before that:what will become of ours relationship with reality, now that what we see and hear is no longer "true" by definition?

Even in our latitudes, at the time of Matteo Renzi's deepfake on Striscia la Notizia, many observers pointed out that the somewhat ambiguous presentation of the content in prime time could give rise to misunderstandings with serious consequences:Wired had it defined “a further challenge for journalism and for public opinion as a whole”.

But abroad, as often happens, the debate on deepfakes is much further ahead than what hits the front pages in Italy:last year for the first time one massive campaign of videos showing fake, unusually pro-Chinese TV actors flooded English-speaking Facebook and Twitter, opening a new and worrying front of theinformation warfare geopolitical.

Who will save us from the invasion of ultra-fakes?

You certainly don't read them, at least for now:the only state to have passed urgent measures to try to stem the phenomenon is the China, which earlier this year made it mandatory to obtain the consent of the "falsified" person and show distinctive signs, such as watermarks, that attest to the fake nature of the work.But the dragon also has the same problems as the rest of the world:How do we stem a phenomenon whose most ferocious abusers often operate anonymously, on fluid platforms where content multiplies so quickly that they can act almost undisturbed?

Something, if nothing else, is starting to move in old Europe too: Artificial Intelligence Act, just approved by the EU Parliament, aims to impose a new international standard for the regulation and transparency of artificial intelligence applications.But many parts of its text are the subject of debate and legal doubts:it is difficult to define the scope of what can be labeled as "artificial intelligence", let alone how difficult it can be to regulate it.For now, in short, those photos and videos of people who don't exist are destined to do good and bad weather:and we, as in that great Sanremo classic, no longer see reality.

How to recognize a deepfake

Faced with all these risks, all that remains is to understand how to defend yourself from deepfakes:which is not at all simple, and will become increasingly less so, with the exponential progress of these (in the latest Photoshop, Adobe presented a generative function capable of creating photomontages that are almost indistinguishable from reality).

At the time of writing, the main artificial intelligences still have some problems with the surrender of the hands of human beings:if you see an upper limb ending that has something strange, don't rule out that it could be a fake.

The Mit recommends, among other things, of pay attention to the skin on your face of the protagonists of suspected deepfakes:if it appears too tight, or vice versa too wrinkled or with different degrees of age in different parts of the face, then there is a good chance that that image, or that video, is not real.And even blinking and lip movement can give good indications in this sense.