👺 Scammers Turn to Deepfakes as Fraud Rates Surge
Since the beginning of 2026, the use of deepfake technologies by cybercriminals has doubled compared to the same period last year.
Concurrently, a tectonic shift in technology has been documented: the share of fraudulent videos generated by neural networks directly from text prompts (text-to-video) has jumped from 17% to 51%.
These findings are detailed in a report by the Center for Social Engineering and Artificial Intelligence Analytics in Cyberfraud (CSAC), operating out of Cyberdom. Since the start of the year, industry specialists at the autonomous non-profit organization (ANO) “Dialog Regioni” have identified over 160 advanced deepfakes, noting that the total number of such generations has multiplied 26-fold over the past two years.
Experts report a radical shift in criminal strategy: perpetrators have largely abandoned aggressive pressure and threats, pivoting instead to trust-building tools. Scammers are cloning voices and generating video messages on a massive scale, impersonating the victim’s relatives, friends, or colleagues, and utilizing specialized dark web chatbots to draft convincing scripts. According to Sergey Maklakov, Head of Department at ANO “Dialog Regioni,” the quality of modern AI models has reached a threshold where generated content can no longer be distinguished from real footage by the naked eye.
In parallel, the first quarter of 2026 saw a surge in phishing attacks via instant messengers (including the Max platform)—scenarios involving account theft through malicious links accounted for 20.8% of all complaints submitted to the Center for Legal Assistance to Citizens in the Digital Environment.
Fraudulent activity remains traditionally tied to the calendar: ahead of national gender-specific holidays, the number of phishing resources jumped 1.8-fold. Alongside this, new schemes have emerged: “drop-calls” requesting the digits of the caller’s number instead of an SMS code, targeted IT attacks on marketplace pickup point employees to process fraudulent returns, and mass mailings regarding “authorization via electronic signature” sent on behalf of government agencies, in which criminals deploy 32 distinct thematic narratives.
Source: RBC
