The expert spoke about the risks of fraudsters using video clips.

Video clips are becoming a fraud tool that poses a threat to both businesses and ordinary users. Attackers can impersonate company executives in order to obtain the transfer of large sums of money, as well as use fake video messages to deceive citizens. At the same time, existing technologies for recognizing artificially created content do not provide one hundred percent protection. Dmitry Anikin, head of data research at Kaspersky Lab, told Izvestia about this.

Exchange for hacking: how the attack on Binance affected the global cryptocurrency market

Hackers stole $6.3 billion worth of assets, triggering the fall of bitcoin and other digital currencies.

According to the expert, the most serious business risks are related to financial transactions that fraudsters are trying to initiate on behalf of management. For private users, the range of threats is much wider — from the spread of misinformation and false investment recommendations to fake video clips in messengers.

"Several major cases of fraud have already been known, when attackers pretended to be directors of companies for the illegitimate transfer of large amounts of money," Anikin said.

Various technologies are used to detect deepfakes today, including machine learning-based classifiers and video file metadata verification. However, the expert emphasized, even automated systems are not able to completely eliminate the possibility of fraud. Therefore, the effectiveness of protection depends not only on the accuracy of algorithms, but also on how they are integrated into the processes of verifying information and conducting financial transactions.

"Automatic verification can help combat the threats of fake videos, but the processes in which this will be implemented are primary. Businesses should have security protocols that make it difficult to make unreliable money transfers, and ordinary users should adhere to the basic rules of cyber hygiene," the specialist explained.

A separate difficulty is the detection of image substitution directly during a video call. According to Anikin, there are no fundamental technological differences between the analysis of the finished recording and the video stream, since in both cases the algorithms process the video data. However, when checking live broadcasts, there are additional restrictions related to the confidentiality of information.

In particular, the deepfake detection system should be integrated directly into the user communication process, which requires special attention to the processing of sensitive data. Therefore, the development of counterfeit recognition technologies should be accompanied by improved information protection mechanisms and internal security procedures.