A new "observer" has appeared on the USE that is almost impossible to deceive – artificial intelligence for the first time independently identified a violator during the exam. The system pointed to a suspicious participant, and after verification, it turned out that he indeed had a gadget with him. This was reported to TASS by the head of Rosobrnadzor, Anzor Muzaev.
The neural network was developed and "grown" for several years. It is integrated directly into the video surveillance system and automatically analyzes the behavior of examinees. Cameras record what is happening, and the program tracks atypical actions of exam participants – attempts to cheat, use a phone, cheat sheets, or micro-earpieces.
When the system detects suspicious behavior, it automatically places a tag. But the final decision is made by a human: all signals from the AI and online observers undergo additional verification by experienced administrators who confirm or reject the fact of a violation.
Muzaev clarified that the technology operates within the "Smotri USE" platform in conjunction with Rostelecom.
The system is now sufficiently "trained": in 2026, it will be five years since its development. During this time, the neural network has analyzed data from approximately 60 thousand classrooms.