Every now and then, while scrolling through your social media feed or exploring the internet, you might run into an image or video clip that raises your eyebrows. It’s showing something or someone you know, but the event it’s depicting…that never happened. The scenario is too familiar, yet too strange. Welcome to the world of ‘deepfakes.’
Deepfakes are a new wave of technological trickery capable of distorting the fabric of digital reality. They can alter people’s appearances, put words in their mouths, and change scenarios as we know them. Some examples would be your favorite celebrity endorsing a product they’ve never even used, or a politician claiming things they’ve never said – the power of deepfakes is disconcerting.
As deepfakes become more realistic and widespread, the potential for misuse escalates. Now, we need to equip ourselves with the knowledge to separate fact from fiction, real from manipulated. That’s why in this blog post, we’re going to discuss how they work, and share techniques to spot and stop them.
Deepfake Illusion: How They Work
The Tech Behind Deepfakes
The term “deepfake” melds “deep learning” with “fake,” hinting at its tech-driven nature. In essence, a deepfake involves digitally altering video or audio to make it appear as though someone is saying or doing something they never did.
Unlike traditional video editing, which might splice segments or overlay audio, deepfakes rely on more sophisticated technology. At their core, deepfakes are powered by Machine Learning (ML) and Artificial Intelligence (AI). These technologies analyze countless images or audio clips of the target person, learning how to mimic their facial expressions, voice, and even mannerisms down to the smallest detail. Then, like a digital puppeteer, they can generate new content where it appears the person is doing or saying something entirely fabricated.
Platforms like DeepFaceLab and tools offered through Google Colab have made the creation of deepfakes more accessible. These tools dissect a person’s facial characteristics from various angles and expressions, then apply this learning to a new video, seamlessly inserting the target’s likeness.
From Funny to Malicious: The Evolution of Deepfakes
Initially, deep fakes were used to produce humorous content like memes, like inserting a well-known actor into a role they never played or merging speeches from different politicians for a laugh. These early, lighthearted experiments showcased the potential of deepfakes to amuse and astonish.
As the technology behind deepfake becomes more advanced, their negative impact is becoming increasingly apparent. Deepfakes are now being utilized for shady purposes such as spreading false information, damaging reputations, and manipulating public opinion and even elections. This ability poses a serious threat to the accuracy and reliability of information in the digital age. It is now important to be able to distinguish between deepfakes and genuine content to know what’s real and avoid spreading fake information.
How To Spot Deepfakes
While the technology behind deepfakes continues to evolve, there are some tell-tale signs you can look for to expose these digital frauds.
- Unnatural Facial Movements: Genuine human conversation involves subtle, flowing facial movements. Deepfakes, on the other hand, often struggle to mimic this accurately, resulting in unusual or jerky facial dynamics. Similarly, the skin texture in deepfakes can occasionally look slightly off—too smooth or too hard.
- Inconsistencies in Lighting or Audio: A deepfake video may exhibit inconsistencies in the lighting between the target face and the rest of the scene, creating an appearance that seems out of place. Similarly, the audio might not match the lip movements, or the tone of voice might not correctly align with the situation.
- Unusual Blinking Patterns: Humans blink naturally, and our blink patterns are surprisingly hard to recreate accurately with AI. So, keep an eye out (pun intended) for odd blinking or lack thereof altogether.
As deepfakes continue to develop, so do methods and tools to detect them. Platforms like Sentinel (TruSTAR technology) leverage AI and machine learning to identify deepfakes by comparing content with a database of real and fake videos. It’s a technological cat-and-mouse game, but with each advance in deepfake creation comes an equal evolution in deepfake detection.
Double-Checking Information Before Sharing
It can happen to any of us. We come across an intriguing video, and only later do we discover it was a deepfake. To minimize falling into this trap, review, and verify before hitting the share button.
As active netizens, we have a responsibility to prevent the spread of misinformation and deepfakes.
Why is this so important?
Firstly, spreading unverified information can lead to misunderstandings and conflict among communities. Secondly, we should protect those who may be less tech-savvy and easily deceived, like our elders. As with any deceptive tool, deepfakes can be used for scams— another essential reason to double-check and verify first, then share later.
Rely on Credible Sources and Fact-Checking Websites
The internet is flooded with websites with various sorts of information, but not all of them are trustworthy. There’s a saying, “Don’t believe everything you read on the internet.” When it comes to deepfakes, this couldn’t be truer.
To navigate safely through the digital world, it’s necessary to depend on credible news sources and fact-checking websites. Websites such as Snopes, FactCheck.org, and others provide a reliable means to verify the legitimacy of content, and they’re super easy to use!
V. Counteract Potential Deepfake Misuse
Safeguard Personal Info: Minimize your digital footprint. Be careful while sharing personal images or videos on public forums or social media. Fewer personal inputs out there means less fodder for potential deepfake creation.
Use Watermarks: If you share personal or professional videos online, adding watermarks can deter misuse. Even if a video is manipulated into a deepfake, the presence of your watermark can signal that the content may have been altered.
Beware of Dubious Links: Do not click on links from unknown sources. Always double-check the link address and verify its safety before clicking. This precaution will protect you from malware that could potentially hijack multimedia content from your device
Use Authentication: Use two-factor or multi-factor authentication for all online accounts. This adds an extra layer of security and ensures that you have control over your content.
Resources for Deepfake Awareness
Here are a few respected news platforms that do an excellent job of breaking it all down:
The New York Times: They’ve got the pulse of the world’s breaks and developments, and deepfakes are no different. Their Tech section provides an in-depth look at the latest in manipulation technology.
BBC: Packed with insightful articles, BBC does more than just news. They dig deep into the ethical concerns surrounding deepfakes and how they’re shaping our society.
Level-Up Your Media Literacy
Fake news, misinformation, now deepfakes. In today’s digital world, media literacy has never been more crucial. Here’s egghead approved websites to level up your media literacy skills:
Media Literacy Now: A powerhouse promoting media literacy education, they provide strategies to foster critical thinking and discernment in today’s tech-driven world.
Common Sense Media: Dedicated to helping families and educators in the maze of media and tech, they offer tips on how to demystify deepfakes for kids (and adults too).
Champions Fighting the Good Fight
Against the landscape of strange and scary tech effects, some stand firm and fight the good fight. Multiple organizations are waging war against deep fakes, including:
DeepTrust Alliance: Working towards a trusted digital environment, DeepTrust Alliance is committed to countering deepfakes and promoting digital trust.
Sensity (previously Deeptrace): They focus on tracking and tracing deepfakes via detection technology, providing a much-needed defense against manipulation.
Playing the Digital Defense
Also, in the thick of this battle is the National Cyber Security Alliance. Devoted to online safety and privacy, they’ve got excellent resources and tips on how to strengthen your digital defense against deepfakes.
There’s no easy solution to the deepfake issue. Staying informed is our best shot. These resources help us understand the threat and our role in combating it.
Conclusion:
Deepfakes, at their core, are powered by Machine Learning and Artificial Intelligence and can convincingly alter video or audio content, skewing reality to the point of being indistinguishable. However, there are ways to spot them. With technologies like Sentinel and others, we also have growing means to combat these digital deceptions.
If you like this article, be part of our community and join us in growing digital literacy, and critical awareness, and nurturing a culture of accountability online.