Over the past few years, you may have heard the term ‘deepfakes’. Despite hearing the term, many people are unsure of what a deepfake actually is and how it could impact society.
This impressive AI technology has the power to manipulate images, video and audio. Despite being impressive, this technology is becoming more advanced and accessible.
- Deepfake tech uses artificial intelligence to create artificial images, video and sound.
- Most deepfakes are used on celebrities and politicians.
- Deepfake tech is becoming more accessible and is being used to harm people’s reputations, create fake news, scam people and more.
- There are positives with this tech, especially for the entertainment industry.
- Tech companies and governments are starting to tackle deepfakes.
What is a Deepfake?
Deepfake technology gives users the ability to create artificial videos, images and audio clips that look and sound like someone else.
The best example of deepfakes are videos in which someone’s face has been digitally altered to make them look like another person.
Using a form of artificial intelligence called Deep Learning, AI has the ability to create fake media. This includes swapping peoples faces in videos, creating completely fake images and even cloning voices.
While face-swapping has been a feature on Snapchat – and other smartphone apps – for many years, the recent advancements in AI technology are making this type of fake media more unrecognisable.
Why and How are Deepfakes Used?
Most deepfakes feature celebrities and politicians. This is because the AI to create deepfakes requires as many images of that person to generate a believable video/image. Hence why deepfakes of celebrities and politicians are more common.
The majority of deepfakes are currently created for porn, resulting in many unsuspecting women – mainly celebrities and politicians – becoming victims of this tech.
Many deepfakes are being used in politics, making politicians saying certain things they never said. The most recent example was a fake video shared by a political party in Belgium that featured Donald Trump telling the country to leave the Paris Climate Agreement.
Deepfakes can also be used to clone peoples voices, creating audio clips of someone saying things they never really said.
For example, a CEO of a UK energy firm thought he was on the phone with his boss, however, he was being scammed by someone using an audio clone of his boss’s voice. He ended up transferring over $200,000 to a Hungarian supplier.
How’s that for a bad day at work?
READ MORE: How VR Will Change Our Lives
How are Deepfakes Created?
Deepfakes are created using Artificial Intelligence. In particular, a branch of AI called Deep Learning. This uses neural networks, a type of machine learning technique similar to how our brains work.
The user first trains a neural network on hours of real video footage of the person, helping the AI understand what that specific person looks like from different angles, in different lighting and in different environments.
They combine this trained network with computer graphics to swap the faces of the specific person onto someone else’s face.
While AI makes this process more efficient, it still takes the creator a lot of time to deal with any obvious glitches in the video.
Why Deep Fakes Are Becoming A Problem:
Deepfake technology was once only available in the likes of VFX rooms at Hollywood studios. Now, this tech is being passed down to anyone with a good PC and knowledge on how to create artificial content. Deepfake technology is becoming easier to use and much more accessible.
While this technology still relies on top of the range computers, the tech will develop over time and it will become increasingly easier to create deep fakes from an average smartphone.
There are already apps for smartphones that have the ability to swap peoples faces and animate peoples faces.
The worst part about all of this is that these deepfakes are becoming more believable and their quality will only improve over time.
Some Examples of Deepfakes:
In 2018, Jordan Peele and Buzzfeed teamed up to make a PSA, warning the public about the effect of deepfakes on society. The video featured a deepfaked Barack Obama calling Donald Trump a ‘dipshit’.
Most recently, a TikTok user started posting Tom Cruise deepfake videos, instantly going viral. The user’s technique was to find someone that looked like Tom Cruise and use AI to put Cruise’s face on the actor.
Problems Deepfakes Have On Society:
There’s a justified concern that deepfakes will have some seriously negative effects around the entire globe:
- Deepfakes could be used to ruin trust and credibility in politicians.
- Deepfakes are already being used for revenge porn.
- Fake evidence could be entered in court trials.
- Phone scams and identity fraud would be easier than ever before.
Deepfakes could lead to a society where we’re unable to trust the things we see online. Any image or video could be completely fake.
The general public is already cautious of misinformation and the things they read online, but imagine a world where we also can’t trust the things we see.
If there was a lack of trust in the media we see online, people – especially politicians – could label real images/videos/audio clips of them doing something illegal as being fake.
It will be much easier for a politician to claim certain images and videos are AI-generated if the public is already cautious of the things they see online.
The Positive Side To Deepfakes:
Despite all of the negatives, there could possibly be some upsides.
Deepfakes can actually be entertaining and sometimes useful.
For example, it could help create virtual humans, characters or avatars for use in video games. With the emergence of Virtual and Augmented Reality, this could be quite useful.
The power of deepfakes could even – somewhat – bring back people from the dead. Imagine being able to hear the voice of a loved one who recently passed, or even reanimating an image of them.
Deepfakes could also be useful for foreign films with actors and actresses seamlessly speaking other languages. A similar example involved David Beckham and a health charity. They used AI to produce a video in which Beckham spoke nine different languages, all with the aim of promoting a campaign against malaria.
While deepfakes do have many negative effects on society, there could be some positives too.
How To Spot A Deepfake:
So we know deepfakes are on their way, but how can we educate ourselves to easily identify them?
At the moment, most deepfakes are fairly obvious. However, some can be more convincing. Here are some tips on how to identify them:
- Look at the eyes, if they look pretty strange and/or not blinking through the entire video, it could be a Deepfake.
- Look for blurs around the face, hair and skin.
- Is the lighting unnatural?
- Check if the audio matches the mouth movement, deepfakes often struggle syncing mouth movement to speech.
How To Stop Dangerous Deepfakes:
Laws tackling deepfakes in the US have already been passed in some states, some criminalising deepfake porn and prohibiting its use during elections.
Companies are also working hard on software to identify deepfakes and watermark them so people are aware the video is fake.
Facebook and Twitter are working on preventing deepfakes from appearing on their platforms.
Ironically, AI is being used to stop AI. Deeptrace and Reality Defender are both examples of software that finds and restricts deepfakes from surfacing online.
It’s Not All Doom and Gloom:
Deepfakes are a serious concern for everyone in society. Moving forward, this technology will get more advanced and fall into the hands of the general public.
To avoid the worst effects, it’s essential we all stay educated on deepfakes, how they work and how we can stop them.
In the bigger picture, more countries and tech giants will have to clamp down on this technology to make sure it doesn’t get out of hand.
While deepfakes could have serious impacts on society, there’s still hope those impacts can be controlled and the technology used for good.