Draft Russian plan proposed AI to detect 70% of targeted VPN services

A draft Russian national plan proposed using artificial intelligence to identify virtual private networks (VPNs), with models analyzing characteristics of 70% of services targeted for censorship. The proposal was reported by Russian newspaper Kommersant. TechRadar says authorities reportedly estimated the system could identify VPNs with 95% effectiveness.

Russia’s Digital Development Ministry said the measure was absent from the plan’s current version. Russia currently uses what it calls “intelligent algorithms” to analyze internet traffic. The government also plans to increase filtering capacity 2.5-fold, to 954 terabits per second by 2030, according to the report.

The proposed AI would analyze connection patterns, including packet size and timing, session duration, traffic ratios and nearby IP addresses. An industry source told Kommersant that automation could speed up this analysis. Other industry sources warned that the system could mistakenly block legitimate business traffic. One source estimated that the current algorithmic error rate is 5% to 6%, and said early AI models could produce many incorrect blocks.

The plan would add an AI layer to Russia’s existing traffic-filtering equipment, known as TSPU, installed on telecom operators’ networks. The report also describes the proposal as part of broader efforts to restrict VPNs and control communications. Separately, a draft ministry order would require smartphones using Russia’s future 5G networks to support Kuznyechik, an encryption algorithm developed with participation from the FSB’s information-security centre and InfoTeKS. The proposed requirement is scheduled to take effect on March 1, 2032.