The pandemic-induced surge in online services has highlighted new trends of identity fraud. According to iDenfy, a digital verification service provider, the share of fraudulent identifications has grown twice since 2019, while the frequency of deepfakes detected increased 12 times.
Based on data from the online authentication service provider iDenfy, the COVID-19 pandemic marked both the rise in digital services usage and the number of attempts to commit various ID frauds. In 2019, approximately 6% of all digital identifications were rejected and marked as fraudulent. After more than a year into the pandemic, over 12% of all identifications are marked as attempts to commit identity fraud.
In 2020, iDenfy saw a 2% rise in attempting to use a fake personal identification document to complete registration. According to the company’s head of verification department, Greta Makareviciute, even though counterfeit IDs remain the most popular method of fraud, the pandemic has shown a steep increase in more advanced ways to bypass online verification algorithms.
“As the COVID-19 spread, we have started seeing more and more attempts to trick our biometrics recognition software. We have encountered people wearing hyperreal masks, trying to film relatives’ faces while they are asleep, and, of course, using deepfakes. Deepfakes alone constituted 6.08% of all identifications in 2020, which is a staggering twelvefold increase from 0.5% in 2019,” claims Makareviciute.
Deepfakes is an artificial intelligence-based technology that helps create lifelike images and videos of any person. They aim to mislead the viewer and, in iDenfy’s case, the identification algorithms. In most cases, the end goal of using deepfakes is to commit financial fraud.
“Most of us have seen the famous deepfakes that look like Tom Cruise and other celebrities, and the similarity is truly terrifying. However, such a professional result is truly costly to achieve. The deepfakes that we stumble across most often are far simpler. Usually, they are awful at winking and maintaining realistic facial expressions.
Moreover, the lighting, shadows, and colours are often off. It is also challenging for AI to recreate teeth, hair and natural reflections of glasses. This is why you will never meet a deep fake with curly hair and a great smile,” says Makareviciute.
According to the company, even though rapid technology improvement and increasing accessibility of deepfakes are cybersecurity threats, companies in the online verification sector are creating powerful tools to fight them.
“Currently, every identification is reviewed and supervised by members of the Verification team. AI-based biometric recognition paired with manual checks ensure that people behind the screen are real humans. As the volume of fraudulent identifications does not seem to be decreasing in 2021, constantly improving our liveness detection algorithms is essential to us. In order to do that, we are closely monitoring how hackers evolve and putting efforts to stay one step ahead,” comments Makareviciute.
iDenfy forecasts that the demand for biometric recognition will grow, and the trend will pose new challenges for the scammers and the companies that fight them. The future of biometrics is in pairing the technology with real-time behaviour analysis, online profile checks and similar measures.
Follow us and Comment on Twitter @TheEE_io