Artificial intelligence technologies allow to automatically identify telephone scammers at the call stage — even before the subscriber has time to answer. Alexander Lunev, head of the information security training group at Yandex, told RIA Novosti about this. According to him, automated systems based on machine learning analyze the number and behavioral patterns of the call, and then mark it as suspicious directly on the smartphone screen. This is especially important against the background of the growing number of attacks using SIM boxes — devices for mass calling that allow scammers to attack hundreds of subscribers in a few minutes. Manual analysis of such calls takes a lot of time, while AI systems make a decision in less than a minute.
How the Algorithm Works
The system takes into account more than 300 behavioral factors, including:
- frequency and duration of calls from one number;
- mass simultaneous calls (typical for SIM boxes);
- who ended the conversation first — the subscriber or the caller;
- anomalies in call routing and number spoofing (spoofing).
Neural networks take into account more than 300 behavioral factors: the frequency and duration of calls, mass simultaneous calls from one number, and even who ended the conversation first. If the system detects an anomaly, such a call is automatically marked as suspicious on the smartphone screen
The algorithms work with anonymized information, user data is not transmitted and is not used for identification. The decision to mark is made on the basis of statistical anomalies in network traffic and the behavior of the calling subscriber. If a number is recognized as fraudulent, a corresponding label appears on the phone screen even before the user answers the call.
Mass Attacks and SIM Boxes
Scammers actively use SIM boxes — specialized devices that allow you to simultaneously call hundreds of subscribers from spoofed numbers. Such devices often operate on the basis of GSM gateways and can be located in any region, hiding the real location of the caller. Yandex AI systems identify patterns characteristic of SIM boxes: short calls, high dialing frequency, simultaneous activity of dozens of numbers from one device. As soon as an anomaly is detected, all calls from this infrastructure receive a warning.
The Human Factor Remains Key
Despite the automation of filtering, the expert emphasized that the final decision always remains with the user. Technical means of protection do not cancel the basic rules of cyber hygiene:
Despite the fact that AI technologies take on the main routine work of filtering mass threats, the final decision always remains with the person, and only by combining modern developments with personal vigilance can a truly reliable defense be built
Why Using AI Against Scammers Is Important
Previously, marking suspicious numbers required manual entry into databases or the use of third-party caller ID services that worked with a delay.
According to the Ministry of Internal Affairs, the number of telephone fraud cases in 2025 increased by 23% compared to the previous year, and the amount of damage exceeded 30 billion rubles. The main tool of attackers is mass calls with number spoofing, which cannot be blocked by traditional methods.
The introduction of AI systems on the side of telecom operators and in applications with caller ID allows to reduce the number of successful attacks, warning the user before he enters into a dialogue. For subscribers, this means an additional barrier to the fraudulent scheme, and for operators, it is a way to reduce the number of false complaints and the load on technical support.
