What are deepfakes and how do you spot them?

File picture: Pexels

File picture: Pexels

Published Sep 19, 2021


No one likes a catfish. Sadly, catfishing is still popular and with that, deepfake photos and videos are on the rise.

What exactly are deepfakes?

Deepfake is similar to Photoshopping, but worse – and the consequences can be disastrous.

Deepfakes use a form of artificial intelligence called deep learning to create images of fake events.

How does it do all that?

This deep-learning system can produce a convincing counterfeit by “studying” photographs and videos of the unlucky person from multiple angles. It will then mimic the person’s behaviour and speech patterns.

It is worth noting that images, video and audio can be deepfaked, so even if it sounds like the person, it might not be because the system has studied the person’s speech patterns.

Have there been cases of deepfakes used to harm people?

Deepfakes can be used in nice and wholesome ways. You can star in your favourite movie as your favourite actress or you even can be a back up dancer for your favourite artist.

However, if used maliciously, deepfakes can be used for attributing false statements to politicians and other public figures and can turn people into pornstars without their consent.

According to a MIT Tech Review report, deepfake technology has been applied on a platform that promises to allow anyone the ability to turn anyone into a pornstar.

There was also a website that uses this machine-learning to instantly turn normal photos of famous AND everyday women into realistic deepfake nudes.

According to media reports, the site has garnered more than 38 million hits since the start of 2021.

In Pennsylvania in the US, a mother allegedly used explicit deepfake photos and videos in a bid to get her teenage daughter’s cheerleading rivals kicked off the team.

She reportedly sent the content to the coach. The pictures and videos showed the members “naked, drinking and smoking”.

Cyberattackers are also making use of deepfakes. Using the technology, attackers can show people or businesses in all sorts of illicit (but fake) behaviours that have the power to damage their reputation if the images went public.

They are then told to pay the ransom if they want the videos to stay private.

So how can you spot them?

Unfortunately, identifying deepfakes gets harder as technology advances. There is hope. Deepfakes that have not been done as well are easy to spot as the lip-synching might be off or the skin tone does not look particularly right.

You might even notice flickering around the edges of the deepfake faces. Because the algorithms sometimes have trouble rendering jewellery and in some cases, teeth, these could be dead give-aways too.

It is important to verify whatever you see. If you receive a message, double or even triple check its source. If you are suspicious about a picture, do an image search to find the original, if possible.

So it might seem that one day we will all fall victim to deepfakes – but maybe not. Governments and big tech firms are funding research to detect deepfakes.

Facebook and Microsoft have taken large strides to detect and remove deepfake videos. In 2019, the two tech firms announced they would be collaborating with top universities across the US to create a large database of fake videos for research, according to a Reuters report.

Cybersecurity teams in various companies worldwide are also working around the clock to prepare defenses as the tools improve.

How do I protect myself against deepfakes?

As we are in the era of deepfakes, it important now more than ever to make attempts at protecting yourself against potential deepfake attack.

When it comes to your own media, use a digital fingerprint or watermark. This will makes it more difficult for someone to create synthetic content and deepfakes from your pictures or videos.