Spotting Deepfakes: Your Guide to Protection & Awareness

For decades, popular science fiction movies like 2001: A Space Odyssey and The Terminator have warned us of the dangers of unregulated AI. However, current AI technology has enormous benefits and can be helpful in every type of busines. AI models like ChatGPT can even help us to become more efficient in our daily lives. Thankfully, although the prophetic AI movies are incredibly entertaining, their doomsday visions haven’t come to fruition. We don’t have to worry about HAL or Arnold Schwarzeneger-like killer robots wiping out humanity. Yet the increasingly sophisticated technology has become a tool that can be dangerous in the wrong hands. Deepfakes are the latest AI-generated output to present a unique and relatively new threat to anyone who accesses the Internet. Let’s take a look at what deepfakes are and the strategies to spot deepfakes and protect yourself. Cybercriminals continually look for ways to weaponize technology against unsuspecting Internet users. Each year, it seems a new threat rears its ugly head and presents new risks to people online. The most recent trends include deepfakes that at best, are used to prank friends, and at worst, spread misinformation and steal confidential data. You may have seen trending, comical AI photos of celebrities with extra limbs and fingers. But deepfakes aren’t always easy to detect. A deepfake is an artificial image, recording, or video that’s produced by Artificial Intelligence (AI) via a process called “ deep learning .” Deep learning teaches AI to process data in the same way the human brain works. Basically, this means that deep learning allows AI to learn through trial and error. It can figure out what works and what doesn’t, and accordingly alter its output. AI trains on existing images, videos, and voices, and uses those to refine the “deepfake” it was told to produce. For example, if you wanted to prank a friend by placing yourself in a tropical getaway, you would feed AI images of yourself and the Maldives to create your video. Deepfakes aren’t only for cybercriminals, scammers , or pranksters either. They allow us to restore lost human voices and historical film footage and images. Although the propensity for evil intent looms large, the technology itself can prove extremely beneficial. The recorded voice sounded like President Joseph R. Biden, but it was a “bunch of malarkey.” In the weeks leading up to the 2024 New Hampshire presidential primary elections, registered Democrats in the state received a recorded call from “Joe Biden.” The caller identified as the current U.S. president, and urged voters to stay away from the primaries. This call even used President Biden’s trademark old school phrase, “what a bunch of malarkey,” to describe the primary vote. However, the caller wasn’t the president. It was a deepfake AI robocall thought to have originated from AI technology made by ElevenLabs. The seemingly innocuous “prank” attempted to commit the 100% illegal act of voter suppression and purposeful dissemination of misinformation . It left the general public worried about increasingly sophisticated AI technology. And it prompted lawmakers and policy advocates to call for heavier AI regulations. Unless you know how prolific deepfakes have become, how to detect deepfake signs, and how to protect yourself from falling victim to deepfakes, it’s easy to fall prey to imposters spreading false statements and news. In recent years, deepfake schemes have made international headlines. Some of the most infamous deepfake examples can help you understand how to spot deepfakes and protect yourself from their malicious intent. Here are two of the most infamous deepfakes: Taylor Swift sparked the ire of armchair quarterbacks everywhere when she began dating Kansas City Chiefs superstar tight end, Travis Kelce. The 30 seconds of camera time she received during Chiefs games proved too much for some random Internet users, and in January 2024, deepfake, compromising photos went viral. According to TechCrunch , one of the deepfake images was viewed over 45 million times on X (formerly known as Twitter). It took days before the images were deleted, and the Telegram group thought to be responsible has yet to face real consequences. In 2017, researchers from the University of Washington experimented with deepfake tools to heavily edit audio clips and an AI-generated image of President Obama to produce a video. The scarily realistic video showcased Obama espousing views on job creation and other important topics … but Obama had never actually said any of the things the video showed him saying. If you’ve bought into a video or audio clip you’ve found online, only to later discover that what you viewed or listened to was a deepfake, you’re not alone. Deepfake technology is constantly evolving, and can be difficult to detect. The creation of a deepfake requires high-end graphics cards and sophisticated cloud computing power. Thus, when used for malicious purposes, deepfakes are typically made by a group of hackers or cybercriminals, rather than by a lone bad actor. It’s vital to be able to distinguish between real audio or video clips and deepfakes — especially as numerous public figures, like Prince Andrew , have cried “deepfake” when presented with evidence of images or videos capturing their actions and words. Some of the signs to detect a deepfake include: It sounds weird, but normal blinking patterns are difficult for AI to correctly generate. If you’re watching an online video and wonder if it’s a deepfake, pay attention to whether or not the individuals featured blink. Deepfakes may not perfectly sync audio to the lip movements of AI-generated images. Even if a clip features a well-known figure, pay attention to the audio to help in spotting deepfakes. Fingers are often a human feature AI seems to have a difficult time recreating. Although it may be tough to distinguish real facial features from those generated by AI, deepfakes often show people with too many fingers. Researchers at MIT created a website, Detect Fakes , to help you distinguish between AI created images and real people. The researchers also suggest looking closely at people’s complexions — does a person’s skin look too smooth or too wrinkly? A weird complexion is a tell-tale sign of a deepfake. One of the most insidious uses of deepfakes comes via phone calls. Phishing schemes and money scams may spoof the phone number of a loved one to ensure you answer the call. The caller will then claim that your loved one is in trouble and needs money. They may ask you to verify confidential information, too. The most sophisticated scams will use an AI-generated voice to mimic your loved one. A good countermeasure is to agree on keywords to share in a conversation that only you and your loved ones would know. Once you understand strategies for spotting deepfakes, you’re better equipped to protect yourself against their designs as well. Here are three strategies for protection from nefarious deepfakes: A British energy company CEO received a call from the company’s president requesting an immediate transfer of funds. The CEO didn’t follow up with his boss via email or another call, and didn’t notice that the transfer ended up in a third party account. The call was a deepfake. If you receive a weird personal or professional request, confirm the identity of your caller via email or an identifier only your real contact would know. Ensure you can discern reality from deepfake by looking for the methods of detection listed above. If a video or a call seems subtly off to you, chances are it’s a deepfake. If you receive a robocall purporting to be from a public figure or loved one, listen to the rise and fall of the voice on the other end. Are there inflections and a change in tone? If not, you may be on the receiving end of a deepfake. Deepfakes can present numerous cybersecurity risks to unsuspecting recipients. If you’re on the receiving end of a deepfake, you may find your personal data and financial assets exposed and vulnerable. You could also digest misinformation and spread falsehoods that influence the choices of others. Now that you understand some strategies for spotting deepfakes and protecting yourself, you’re better equipped to avoid the cybersecurity risks associated with the growing AI trend. Visit What Is My IP Address to utilize our cybersecurity tools. And be sure to check out our blog for the latest insights and tips to protect yourself online.