- Deepfakes increasingly trick users into committing fraud and scams
- AI-powered attacks now involved in 16% of breaches
- Detecting fakes requires observing facial/audio glitches and urgent requests for money
Deepfakes, technology that uses artificial intelligence (AI) to synthesize realistic audio, video, and images of real people, have improved to the point that even attentive people can be fooled relatively easily.
By creating multimedia that showed celebrities, politicians, or tech leaders saying things they didn’t say or doing things they never did, Internet fraudsters got people to authorize fraudulent transfers or innocent people to “invest” in fake projects on fake investment platforms. In some cases, they even managed to create a fake emergency (such as a car accident or attempted kidnapping), which sent family members scrambling to make payments.
Today, 16% of all breaches involve AI-powered attacks (according to IBM data), with phishing and deepfakes being one of the most commonly cited methods, says Danny Mitchell, cybersecurity writer at Heimdal Security, and after delving into the growing scam technique, he discovered that there are still ways to detect a deepfake and has shared his findings with TechRadar Pro.
Article continues below.
Four practical steps to detect a deepfake
According to Mitchell, one of the ways to detect a deepfake is to observe unnatural facial movements or blinks.
“Deepfake videos can struggle with the subtle mechanics of human expression,” Mitchell writes. “Watch for blurred facial edges, inconsistent blinks, or expressions that don’t match the emotion being conveyed.”
He also maintains that the audio can sometimes be slightly off in deepfake videos. AI-generated voices may have a slight monotony or unusual rhythm, while background noise may appear artificial.
Additionally, deepfake videos often have mismatched lip movements, as the synchronization between speech and lips is often imperfect. This is particularly visible at faster talk speeds.
Therefore, detecting deepfakes comes down to searching for pixels. But there are other ways to identify a fake, and that is to take into account the content of the message. Mitchell says that urgent requests for money or sensitive information, which put pressure on users to act quickly, are the common red flag that users should pay attention to.
The best antivirus for all budgets
Follow TechRadar on Google News and add us as a preferred source to receive news, reviews and opinions from our experts in your feeds. Be sure to click the Follow button!
And of course you can also follow TechRadar on TikTok for news, reviews, unboxings in video form and receive regular updates from us on WhatsApp also.
