There is a popular saying that seeing is believing. With so many fake videos circulating online, it’s hard to tell real from fake.
So, is seeing truly believing?
In this article, we will look closely into the controversial topic of deepfake technology.
What is Deepfake Technology
Deepfake technology is the use of artificial intelligence and machine learning to alter videos,
images, or audio recordings to look and sound just like a real human. Previously, it was mostly
used for gimmicks, to mock celebrities on social media. Just for fun. But it’s getting our attention
now, as it has become more sophisticated. Over the past few years, the growth in deepfake
material has been rapid. Indeed, the deepfake technology market is massive. As of 2024, it has a
market value of £6.26 billion and is expected to reach £38.44 billion by 2032. Much like any other
tech innovation, it’s good to weigh the cons against the pros.
Deepfake Technology Cons
According to a recent survey, it’s projected that by 2024, approximately 95% of U.S. consumers
will have been affected negatively by a deepfake. In a recent experiment by Dr. Sophie
Nightingale and Professor Hany Farid at the World Economic Forum asked a group of participants to identify real and fake faces. The majority of the participants voted the AI-generated faces as more trustworthy than the real faces. As deepfake technology gets more sophisticated, it is becoming harder to differentiate real from fake, unless with a special deepfake detection software.
A deepfake can be created using two different programs, an encoder that creates a replica of a
target person from an image or video, and a decoder that compares with the original image/video
until it’s unable to differentiate the two.
In 2020, a British news broadcast service, channel 4 aired a video of Queen Elizabeth making a
Christmas speech, and People tuned in, believing it to be real.
However, towards the end of the broadcast, she stood up and performed a TikTok dance
routine—that’s when viewers realized it was a fake.
Indeed, financial fraud is the main target of deepfake technology by cybercriminals, because of
its monetary gain. In February 2024, some cyber criminals duped a bank employee in Hong Kong
and received $25 million from him. The fraudsters used deepfake technology so that he would
think he was having a real video call with the chief financial officer(CFO). Then they asked him
for some money. He thought this call was real, but in fact, it was fake. That’s how deepfake
technology can create an illusion.
Another serious downside of deepfakes is the risk of reputational damage. Creating a deepfake is
not rocket science. Even an average person can make one with a simple Google search.
Deepfakes can be used to create lies and deceits about a person. The resemblance will be so
similar that it’s hard to distinguish. In 2020, the BBC reported about a woman who used
deepfakes to create a hyperrealistic video about her daughter’s rival in the cheerleading team. In
this video that went viral, the girl could be seen naked, vaping, and singing. The video was
believed by many and caused damage to the cheerleader’s reputation.
But deepfakes aren’t only used for bad. Businesses can also leverage it for good.
Deepfake Technology Pros
The fake in ‘deepfake’ makes it seem as though there’s nothing positive about it. However, there
are several positives in deepfake technology. In 2015, Hollywood utilized deepfake technology in
the movie Fast and Furious 7 upon the demise of lead actor, Paul Walker. In essence, deepfake
technology was used to bring Paul Walker back to life for the audience’s entertainment.
Deepfakes can also be used for educational purposes. For example, employing historical figures
like Albert Einstein to teach students on certain topics. Customer service can also be simulated
by an organization for enhanced customer experience.
Deepfakes can also be used in content creation. For instance, Advertisement. Cara Delevingne is
a good example of using deepfakes in an ad campaign. A 2018 Zalando’s #whereeveryouare
campaign featured deepfakes of supermodel Cara Delevingne, allowing the creation of 290,000
localized ads for towns and villages across Europe. The supermodel didn’t need to be physically
present for extensive shoots.
How to Spot Deepfakes
Spotting deepfakes is becoming increasingly difficult without the help of emerging technologies.
However, below are tips for spotting these deepfakes
- Blinking. Look closely at the eyes. Is the person blinking too much or too little? This will give
it away. - Facial hair. Does the facial hair appear weird? Deepfakes may make facial hair appear too full
or scanty. - Lips. Deepfakes may make the lips appear different from the face. That is, the size and color of
the lips do not match the face. - Body movements. Pay close attention to the body movement. Does it appear robotic and
animated? - Face. Zoom in and look at the face. If the face looks weird, for instance, the ears, mouth,
forehead, and lips. It appears unnaturally symmetrical. - Shadows. Look for unusual shadows that do not fit in with the lighting.
- Lip Synch. The Lip synching is usually not in synch with the original. It either talks too fast or
too slow. - Audio. When it comes to audio, if the voice doesn’t sound human. If it sounds robotic.
- It is fake. If it appears fake, then it is probably a deepfake.
Mitigation strategies
These are measures that can be put in place, to mitigate the negative effects of deepfakes;
- Technological Tools
Detecting deepfakes manually may be exhausting so the help of AI deepfake tools, makes
the work easier. - Multifactor Authentication
Implementing multifactor authentication in your accounts is very important. This is when extra
steps are needed to have access to your account such as asking personal questions or using an
authenticator app. - Watermark Videos/Photos
When sharing photos and videos online, employ safe measures by watermarking them so it’s
difficult for a third party to manipulate these for deepfakes. - Continuous Learning
The best way to stay abreast of deepfake technology and AI is to keep learning. When users and
employers read news and trends, they’ll stay current on the happenings in the AI world and thus
will have ideas on how to tackle malicious deepfakes. - Policy and Legislation
There is a need to put policies in place that will guide the use of deepfakes, and also ensure
enforcement with legislation. This will ensure scrutinization and prevent unauthorized use of
media.
What Can Be Done
While taking advantage of the good side of deepfakes is beneficial, the bad side also has to be
weighed.
As such, staying informed and training oneself on how to spot deepfake videos/images is key to
not falling prey to vicious deepfake attacks.
