- Статьи
- Economy
- Caught in a neural network: the number of deepfakes in Russia has increased 10-fold in 2.5 years
Caught in a neural network: the number of deepfakes in Russia has increased 10-fold in 2.5 years
In Russia and the world, the number of unique deepfakes is rapidly growing — cases where intruders create a realistic digital double of a person for a specific purpose. In 2.5 years, the number of such cases in our country has increased 10.7 times, VisionLabs told Izvestia. The latest high—profile cases are related to the substitution of governors - they allegedly warn in video clips about rising housing and communal services tariffs, rising food prices, or turn to foreign airlines. Experts expect that the number of deepfakes will continue to grow, and the tools for creating them will become more accessible. At the same time, ordinary people are increasingly becoming victims of fraudsters, and it is extremely difficult to identify the attackers.
A wave of deepfakes has covered the regions
By the end of 2024, 84 unique deepfakes were identified in Russia, in 2025 their number increased to 627, and by September 2026 it reached about 900, Dmitry Markov, CEO of VisionLabs, told Izvestia. According to him, each of them is used repeatedly, which leads to tens of thousands of attacks annually.
Traditionally, the targets are politicians, celebrities, executives and owners of large businesses. This fall, video messages allegedly recorded by the heads of Russian regions began to actively spread on social networks. They address security issues, rising prices, and other sensitive topics.
So, in September, a fake video appeared on behalf of Ilya Seredyuk, Governor of the Kemerovo Region. In the video, the deepfake of the head of the region announced a sharp increase in prices for basic foodstuffs due to an increase in the VAT rate. Later, the press service of the Kuzbass government reported that the video was a fake, there was no shortage of food in the region, and there were no plans to increase tariffs.
Several fake publications of the same type on behalf of regional leaders concerned flights. In one of the recent videos, the deepfake of the governor of the Smolensk Region, Vasily Anokhin, called on foreign airlines operating flights in Russian airspace to "take care of their crews." In his channel in the Maks messenger, the head of the region said that the fake video was created using modern technologies. Similar messages were distributed on behalf of other heads of regions.
Is it possible to steal a voice in three seconds
The increase in the use of deepfakes by scammers is largely due to the low entry threshold, experts interviewed by Izvestia believe. Anastasia Osipova, an analyst at the Positive Technologies research group, said that previously, creating even relatively low-quality fakes required powerful equipment, knowledge of programming languages, and long-term configuration of generation parameters. Now, it's enough to use a ready-made application or online service to create a deepfake for free in a few minutes.
In particular, voice cloning technologies have become widespread. Popular applications only need 3 to 30 seconds of recording time to reproduce the unique characteristics of a person's voice: timbre, pitch, and intonation. In this case, it is not the length of the fragment that is important, but the quality of the recording.
— That is why there are so many public figures among the famous victims of deepfakes: celebrities, top managers of large companies, politicians and bloggers, whose speeches and interviews are published in large numbers on the Internet, — Anastasia Osipova noted.
How not to become a deepfake
Recently, ordinary people who have enough photos and videos on social networks have become victims of intruders more and more often, Anastasia Osipova said. One of the most common motives in such cases is the creation of intimate content without a person's consent. The goal may be financial gain, blackmail, or self-assertion.
— According to various estimates, historically, the vast majority of deepfakes available on the Internet are just such content. Women and girls predominate among the victims: celebrities, former partners, colleagues, and even teenagers," the expert noted.
In fact, to ensure that you don't become a deepfake, you don't need to use the Internet. In modern realities, this is impossible, but it can complicate the task for an attacker, the expert explained. First, to reduce the amount of personal information in the public domain: for example, to close a profile or limit the circle of users who have access to photos and videos. Secondly, it is recommended to arrange a code word or security question with your loved ones in advance. It is also important to enable two-factor authentication, use unique passwords, and monitor the security of cloud storage.
If a fake has already appeared online, first of all it is necessary to preserve the evidence.: screenshots, links, publication dates, account names, and correspondence if blackmail is involved. Then you should complain through the platform or online service where the malicious content is found. As a rule, such sites have special forms for complaints about synthetic materials published without a person's consent.
Businesses are also trying to combat deepfakes by introducing special detectors that evaluate whether it is a live image or a generated one during video calls and conferences, Dmitry Markov noted. According to him, the VisionLabs program assigns an image a score (score) reflecting the probability of authenticity. The closer the indicator is to 100%, the higher the probability that there is a real person in front of the camera.
— The further scenario depends entirely on the customer. He receives a notification about the deepfake and can contact law enforcement agencies or remove the participant from the video call," he explained.
Measures to combat malicious videos are also being worked out at the legislative level. In July 2026, State Duma deputies drafted amendments to Article 63 of the Criminal Code of the Russian Federation ("Circumstances aggravating a crime"). According to the initiative, creating a deepfake in itself does not amount to a crime, but can be considered an aggravating circumstance when committing another offense, such as fraud. However, the Supreme Court of the Russian Federation later pointed out that this issue could not be resolved simply by including the relevant provision in the list of aggravating circumstances: the initiative required careful consideration with the participation of interested departments.
How will the fight against deepfakes change
Experts agree that the number of deepfakes will continue to grow both in Russia and in the world. According to Umar Gebenov, CICADA8's lead data engineer, recognizing the use of a deepfake as an aggravating circumstance only helps to deal with the consequences of an already committed crime. To counter this technology, prevention, detection and prompt removal of dangerous content are necessary, the expert believes.
— In the near future, deepfakes will be more actively used on phishing sites, where neural networks generate unique videos and clone the voices of loved ones or supervisors. Such schemes are already being used, but over time there will be significantly more of them," he noted.
According to Anastasia Osipova, large companies and banks will have to review their identity verification procedures. Soon, it will be impossible to unconditionally trust either the image of a person's face or the voice of a person, especially if the verification is based on only one of these signs. Therefore, organizations will begin to systematically use alternative communication channels: to confirm large financial transfers in an additional way, introduce multi-stage coordination of transactions, and review remote identification procedures using video and biometrics.
According to Dmitry Markov, 73% of people are unable to distinguish a real face in a video from a deepfake. Therefore, it is important for organizations to implement protection tools against such attacks, and for developers to regularly update anti—fraud systems so that they keep up with the improving fraudulent schemes and, if possible, outstrip their development.
Переведено сервисом «Яндекс Переводчик»
