How AI will help in the fight against deepfakes

For the first time, experts have assessed the share of deepfake technology used by fraudsters in the financial market. According to a survey of the financial marketplace “Vyberu.ru” based on the results of 2023, in 3% of cases, fraudsters managed to lure money from citizens by using vote forgery. Moreover, these were often the voices of leaders or relatives. In 2022, victims did not report such a scheme, according to the survey data. Cybersecurity experts have already warned about the intensification of scams that use deepfake voices as a new trend, including on the Russian market (see Kommersant on January 25).

At the “Cybersecurity in Finance” forum, VTB Chairman of the Board Vadim Kulik confirmed that the bank is already observing cases of fraud involving the substitution of biometrics and voices, including through deepfake. In particular, VTB talked about a scheme where scammers use voice messages from hacked accounts to “earn the trust of the victim’s friends and get money.” Sberbank also observes similar cases, says Stanislav Kuznetsov, deputy chairman of the board of the credit institution: they “are isolated in nature, but are a real threat.”

According to a study by Positive Technologies, the first half of 2023 relative to the second half of 2022 saw a significant increase in deepfake attacks worldwide. In particular, in the UK the growth was 84%, in the USA – 250%, in Germany and Italy – more than 300%, in France – 500%. The growing popularity of this type of fraud is explained by the “availability and variability of malicious AI tools in the cyber services market.”

To create a deepfake, you no longer have to be a great specialist and an experienced hacker: according to experts, “you can get by with good quality videos, especially if we are talking about famous people.” It is inexpensive and does not require a powerful computer.

Cybersecurity market participants say that to combat deepfakes it is necessary to “inform people and use artificial intelligence.” However, the first method, based on the extensive experience of bankers trying to fight fraudsters, looks almost hopeless. Despite the growing financial literacy of the population, citizens are losing more and more money. But artificial intelligence, experts assure, “has already reached the level where it is capable of fighting deepfakes.”

The only question is which of the banks, other than the largest, can afford such expensive technology. So far, Vadim Kulik admitted, credit institutions are “not at all ready” for fraudulent attacks against clients using biometrics and deepfake.