Deepfakes: What and why are they so dangerous?

Jorge Felix Last updated: December 14, 2022 Read time: 10 minutes Disclosure

Deepfakes pose great entertainment opportunities but also criminal activities. This article explains the risks and how to avoid them.

Sneak peek at Deepfakes and their risks

Deepfakes are the latest development in AI-assisted multimedia. The means to produce deepfakes are becoming increasingly available to the general public. That involves risks and potential for criminal activities like we’ve never seen before. The only trick to avoid being fooled by deepfake is to detect it by knowing what deepfake is all about.

Perfect replication has been a holy grail of sorts for Sci-Fi for decades. Think about Star Trek’s replicator devices which can reproduce anything from food to tools, except for latinum, which is why it’s the galaxy’s currency. Or the T-1000 robot in the Terminator franchise can adapt its shape to look like any person. Well, digital technology’s current state of the art is getting close to perfect replication, at least when it comes to digital media. So now we have a new technological item, the “deepfake .”It’s a piece of media altered to look and sound like anybody, dead or alive. So the future is here, and it’s not always good news.

What is a Deepfakes?

Deepfake is a piece of digital media. Videos are the most popular kind that is engineered to look genuine.

Deepfakes come courtesy of our most advanced artificial intelligence and machine learning technologies. These cutting-edge tools allow the manipulation of video and audio that come close to the real thing. Very close indeed.

Look at this example below. It features a false Mark Zuckerberg (of Facebook fame) in which he uttered a few words worthy of a Bond villain — even referencing Spectre.

The Washington Post shared this video in 2019 while highlighting Facebook’s act of retaining falsified deepfake videos of House Speaker Nancy Pelosi.

This one still looks a little rough on the edges, so you can tell it’s fake if you look carefully enough. But improvements are inevitable.

Making a deepfake is still a challenging task. First, the AI needs a previously trained neural network, and the training process needs enormous datasets comprised of thousands of images. Then, you need specialized software. After that, you need to do the same with the voice, which is much easier, and then mix the video and the audio so that the lips and the voice are synced.

Who is using deepfakes?

You’d be amazed at the number of deepfakes circulating the internet since they’re so hard to make. But Facebook was able to collect about 1,00,000 of them in June 2020. The social network giant used the collection to develop and train an algorithm to detect deepfakes.

The large number of deepfakes already around suggests that they are becoming increasingly more accessible and cheaper. So as more people become able to make them, their uses will also increase for both good and bad things. So let’s see how they are being used so far.

1. Fake news

Ironically, in our “information age,” fake news is disruptive due to its prevalence and influence over the public. Deepfakes help them make them more credible than ever, so deepfakes are the perfect propaganda tool. They promise to be the ultimate voting manipulation device soon.

There was concern among specialists over the 2020 election. The fear was that deepfakes could swing the election; however, it didn’t happen. But that was not due to a lack of attempts, but because two years ago, the available technology was not advanced enough to develop deepfakes that looked genuine. However, this will change soon enough, and we could find an election in which fake news based on deepfakes could make all the difference.

2. Cybercrime

New technology always brings about new solutions — and that includes crime. The recent trend in deepfakes could bring online criminality to a new level.

Blackmail is one of the cybercriminal’s most popular devices, and deepfakes can help them force people to do whatever they want. Imagine all the cybercrime types you already know, from phishing to sextortion. Then add the deepfake factor in. It will charge them with digital steroids and make them scarier and more effective than ever.

Then, consider that this technology will become more available as time advances. So it will become a standard tool for social engineering (which is already one of the most effective cybercriminal techniques). Social engineering relies on human emotion as the weakest link in the digital security chain, manipulating a person’s behavior to make them do something they would usually avoid.