Beeline Launches AI Model “Kibervnuk” to Protect Teenagers From Scammers
To combat phone fraud, Beeline has launched the Kibervnuk (Cyber Grandson) AI model, also known as Kiberpatsan (Cyber Lad). The service is designed to protect teenagers. The model runs on the operator's infrastructure and complements the previously introduced Kiberbabushka (Cyber Grandma) system.
The AI answers suspicious calls, imitates the voice and conversational style of a real subscriber of a specific age and gender, and uses that realism to keep fraudsters engaged. Since the beginning of 2026, Kiberbabushka has spent roughly 70 hours interacting with scammers. Individual conversations have lasted as long as 30 minutes. On average, about 500 such calls are redirected to the system each month.
A Specialized Cybersecurity Tool
Kibervnuk illustrates how telecom providers are moving beyond passive spam filtering toward active disruption of fraud operations using generative AI and voice technologies. It also provides a clear example of vertical AI, combining a large language model, speech synthesis, fraud analytics, and telecom infrastructure within a single use case. Importantly, the model functions not as a generic chatbot, but as a specialized cybersecurity tool.
The system is expected to reduce the risk of interactions with phone scammers, particularly among vulnerable groups such as teenagers and families with children, where attack volumes have been rising. The technology could also lower the efficiency of fraudulent call centers. Some calls will be handled by AI personas instead of real people, causing scammers to waste time and resources. The initiative strengthens broader anti-fraud efforts and creates a potentially exportable model of AI-powered decoys for combating phone-based crime.
Building a “Cyber Family”
Systems of this kind have significant potential within Russia. Kibervnuk addresses the growing challenge of phone and financial fraud. Banking-sector anti-fraud platforms already prevent large numbers of suspicious transactions. The most likely path forward is the creation of an entire “cyber family,” where a scammer reaches a virtual counterpart matched to the age and gender profile of the phone's owner. That increases the credibility of the conversation and extends the amount of time fraudsters spend pursuing a dead end.
The product could evolve through a broader range of AI personas tailored to different risk groups, deeper integration with banking and telecom anti-fraud platforms, more sophisticated real-time analysis of fraud scenarios and speech patterns, and potentially the automated sharing of information about suspicious phone numbers with government agencies.
Strengthening Regulation
Platform-based anti-fraud capabilities among Russian telecom operators began taking shape in 2023, when Roskomnadzor launched a unified system for combating scam calls made through spoofed phone numbers. All operators were required to connect to the platform. That created the foundation for automated blocking of such calls. In 2025, Russia adopted Federal Law No. 41-FZ, introducing a package of measures targeting phone fraud. The law included mandatory caller identification labels for organizations and sole proprietors, and in August the Russian government approved the rules governing that labeling system. Against that backdrop, Beeline introduced the Kiberbabushka AI model in November 2025, enabling near-instant voice responses that mimic an elderly customer.
Teenagers soon emerged as a distinct high-risk group. In 2025, Russia's Ministry of Internal Affairs reported that the number of minors affected by cyber fraud had increased by roughly 25%. Authorities also recorded more cases involving the recruitment of minors into criminal schemes through phone manipulation and money-mule operations. Against that backdrop, regulation tightened further in 2026. Mintsifry is preparing a third package of anti-cyberfraud measures, known as Antifrod 3. The initiative focuses on closer coordination among telecom operators, banks, digital platforms, and law enforcement agencies.
A New Benchmark in Anti-Fraud
Kibervnuk demonstrates a shift from simple filtering of suspicious calls to active AI-based defense, where the system does more than identify threats - it engages directly with fraudsters and disrupts their operations.
The project's value lies in providing additional protection for vulnerable groups that criminals may target to obtain verification codes, persuade users to install malicious applications, or recruit participants into money-mule schemes. The model can be viewed as a new benchmark for anti-fraud operations. If it proves effective, similar AI conversational agents could appear across other telecom operators, banks, and digital platforms. The next stage may involve moving from individual AI characters to dynamic AI-profile systems tailored to different subscriber categories and fraud scenarios.
As Kibervnuk demonstrates, telecom companies are becoming full participants in the cybersecurity ecosystem. Over the next several years, cooperation among telecom operators, banks, government platforms, and technology companies is expected to deepen.
