The dangers synthetic media pose to your business

Army green T-shirt, beard, serious look: The president of Ukraine, Volodymyr Zelenskiy, appears on-screen and addresses his armed forces. He has a striking message: "This battle is not worth dying for", he says. "Put down your weapons." This video on Facebook by Zelenskiy went viral all over the world in March 2022. Almost immediately came the warning that the video was fake. And indeed, anyone looking closely could see that. This Zelenskiy’s head is too big for his body. And his voice also sounds strange.

Created with software

Like this one of Zelenskiy, fake videos are an example of synthetic media (also known as AI-generated media): movies, photos, texts, and sound clips created or edited with the help of clever software. The technology behind synthetic media is developing extremely fast. A few years ago, a deep-fake video, as a synthetic video is also called, was very special. But now there are dozens of apps and platforms that create unique images, text, and videos for you in seconds. "Some experts estimate that, in 5 years’ time, 90% of what we see online will be synthetic", says researcher Piia Varis. Varis has done a lot of research for Tilburg University on the impact of technology on society.

Real or fake?

Synthetic media have advantages for businesses: you can use an app to quickly create an image for your site, or write blog text. They also have drawbacks, as Varis explains. The biggest problem: "When you can no longer tell the difference between a fake photo and a real one, it can be dangerous. Synthetic media throw the basis of our society off balance: What is fact, and what is not?"

That false information affects all kinds of situations. Ukraine’s secret services were able to warn the army in time that the Zelenskiy video was fake. But in the future, they may not be able to. Politicians worry about the dangers of synthetic media to society: the Lower House of the Dutch Parliament wants deep-fake technology partially banned. The EU is also urging tech companies to take action against deep fakes.

Damaged bumper

Because synthetic media are so easy to create, criminals use them too. An American magazine reports that scammers are using synthetic images from the app DALL-E for insurance fraud. This is how it works: If you enter the words 'Damaged Bumper Jeep Grand Cherokee' into the app, it creates dozens of images of damaged cars in seconds. Fraudsters send such an image along with a claim form to the insurer, and get reimbursed for the 'damage'.

Fraud with a fake voice

Criminals also use synthetic voices to scam companies and their customers, so far mainly in the financial sector. In early 2020, a bank manager in Hong Kong got a phone call from a senior boss whom he knew. The boss asked him to transfer money for an acquisition. When the bank manager did so, it turned out that criminals had faked the boss' voice with clever software, and that he had transferred the money to those criminals. In September 2021, another gang made a fake video of a well-known Russian bank director. He 'promised' his customers a huge bonus. Those who clicked through came to a fake website made by the criminals, where they had to enter their personal details.

Deep learning

How do you actually create synthetic media? The software behind synthetic media uses what is known as deep learning. This is a form of artificial intelligence that mimics the human brain. The result is a 'robot' that learns from big data. The ChatGPT software is trained with text. The DALL-E app was shown billions of images of information. This allows the software to turn text into an image in seconds. If you give the command 'chair and avocado', the software will create a unique image of an avocado chair.

No technical solution

There is no good technical solution yet for recognising synthetic media. "Facebook has tried to create technology that recognises deep fakes", Varis says. "But the technology behind synthetic media is evolving so fast that the technology for recognising or preventing them keeps falling behind (nu.nl, in Dutch)”. Some experts suggest putting a disclaimer or a warning on media created with artificial intelligence. "But that is complicated, because you cannot check if media without a disclaimer is genuine."

Reducing risks

So as long as there is no technical solution for recognising synthetic media, it is especially important to take measures to protect your business from the risks involved. Varis has these tips:

Ensure that procedures are clear

"When people are busy, they think: my boss sounded a bit weird today, but I just do what he asks. Organise your workflow in such a way that it is impossible to get money transferred based on a single phone call."

Create a crisis plan

What if a competitor spreads lies about you on social media with the help of synthetic media? Or if your business is otherwise dealing with fake images or texts? "You need a plan, because speed is very important when it comes to limiting reputational damage. Whom should you get in touch with to have a fake message taken down from a social-media platform? If you already know that, you do not have to spend days looking for a way out." In a crisis plan, for example, you put important contacts and standard messages that you can quickly send in case of an emergency.

File a report

Have you been scammed or threatened on social media? File a report with the police. Also report fraud to Fraud Help Desk (in Dutch), and they will alert other businesses.