Although deepfake technology (the stuff that we use to make memes, swap faces with celebrities, insert ourselves in movie scenes, and bring dead people to life) has been around for many years, what’s different now is that deepfakes are becoming extremely realistic—to a point where an untrained eye or ear cannot discern what is fake and what is real. For example, take a look at these celeb deepfake videos of Tom Cruise and Morgan Freeman. Now just imagine receiving a phone call that sounds exactly like a work colleague asking you to urgently transfer money. Scary, right? This is exactly what’s brewing in the world of cybercrime.
Below are five real-world scenarios where deepfakes can be used to deceive people and launch highly sophisticated social engineering attacks.
1. DIGITAL EXTORTION
Per a recent FBI advisory, a new sextortion scam is emerging where cybercriminals scrape photos and videos of non-consenting adults from places like dating websites, online video chats, and Facebook, and blackmail the victim using deepfake, sexualized images and videos.
While this tactic may sound like an old page from the crime playbook, the exploitation of sophisticated AI-generated deepfake technology is troubling. What if a cybercriminal creates objectionable or questionable deepfakes of an employee and extorts them to share credentials, steal intellectual property, or allow unauthorized access into restricted areas?
2. VISHING
Hackers are increasingly gravitating towards a more hybrid form of phishing where they use a combination of email and voice-based phishing (or vishing) to breach corporate networks. Traditional vishing attacks involved attackers posing as tech support, customer support, or similar because it was difficult to impersonate someone the target knew personally. All this is set to change with deepfake voice generators.
With just three seconds of recorded speech, attackers can now clone and synthesize voices that are almost indistinguishable from their original audio samples. Imagine a situation where attackers download speeches of C-level executives from corporate social media handles, clone the audio, and use it to conduct highly targeted social engineering frauds.
3. ZOOMBOMBING
Zoombombing is a type of harassment where an uninvited guest imposes themselves in an online meeting with the aim to either hijack it, spy on it, or disrupt it. But what if an attacker carefully plants himself into a meeting, impersonating an authorized and trusted person?
Last year, scammers created a fake hologram of a Binance executive on a Zoom call and tricked crypto developers and investors. The FBI claims deepfakes are increasingly being used by cybercriminals to attend job interviews for home office positions.
4. BIOMETRIC ATTACKS
It’s quite common for businesses to deploy biometric-based authentication systems (face recognition or voice recognition systems) to verify employee identities. Deepfake technology is at a stage where it can trick facial and voice recognition systems.
In May 2023, Binance revealed that crypto fraudsters were using deepfakes to fool the crypto exchange’s customer verification (KYC) processes even though the verification involved the user blinking their eyes or looking up or looking down. Binance admitted the deepfakes they encountered were advanced enough to understand and execute those commands.
5. BUSINESS COMPROMISE
In China, a scammer used face-swapping technology to impersonate a friend on a video call and steal nearly a million dollars. Similarly, an attacker can impersonate a business associate, a vendor, a supplier, a business partner, or a C-level executive and make payment requests, order wire transfers, request changes to bank information, or ask for invoice and billing addresses to be updated. Business email compromise (BEC) (a.k.a. CEO fraud) is already one of the costliest forms of cybercrime and with the emergence of deepfakes, BEC scams are likely to become even more financially damaging.
WHAT CAN FAST COMPANIES DO TO MITIGATE THE RISKS OF DEEPFAKES?
Since there isn’t a single system or technology that can determine a deepfake, organizations must educate employees and their ecosystem of customers, partners, and suppliers to be vigilant, to trust but verify, and to think before reacting. Organizations must train employees on the basics of deepfakes: what they entail, how they’re created, how they are used for malicious purposes, how they can damage the business, and how employees can help. Security teams must also run phishing and deepfake simulations so that employees build an instinct to recognize them. Employees must learn to evaluate lip movements, accents, background jitters, alignment issues, or unusual timing, and report any signs of suspicious activities.
It is estimated that in 10 years, 90% of all content on the internet will be AI-generated. When this happens, make no mistake, deepfakes will disrupt society. Deepfakes are already in high demand on underground marketplaces. Until detection technologies, regulations, and law enforcement can catch up, it is incumbent on organizations to educate employees on these threats and develop human-centric countermeasures to mitigate the risks of deepfakes and other advanced forms of social engineering.
Stu Sjouwerman is the Founder and CEO of KnowBe4 Inc., the world’s largest Security Awareness Training and Simulated Phishing platform.