Russia turns to AI filters to hunt down VPNs in 2.2-billion-ruble crackdown
Reported by TechRadar1 source ↓
TechRadar reports that Russia has deployed machine-learning filters to identify and block VPN traffic, with government documents putting the project’s cost at 2.27 billion rubles.
Key points
- The filters are integrated into Russia’s centralized system for controlling internet traffic
- They analyze connection metadata, including packet sizes, timing and server patterns, rather than decrypting content
- A digital rights expert told TechRadar the system has 101 users and helps officials decide what to block
- Telecom sources cited by Kommersant report misclassification rates of 5% to 6% for traffic analysis algorithms
Details
TechRadar reports that procurement documents reviewed by Forbes Russia show 2.27 billion rubles was spent integrating the filters into ASBI, Russia’s centralized system for managing deep-packet-inspection equipment at internet providers. The outlet says Roskomnadzor has deployed the system on its internal infrastructure.
The filters use a predictive model to examine connection metadata, such as packet sizes, transmission intervals, session lengths and hosting-server patterns. TechRadar reports that the model does not decrypt users’ content. Digital rights expert Dmitry Kuteynikov told the outlet that it is classified as “intelligent decision support” and has 101 users; he said it helps staff decide what to block.
Kuteynikov and a developer for Russian VPN provider Amnezia told TechRadar that traffic-pattern analysis could affect a range of VPN protocols, including obfuscated connections. The article also reports, citing Kommersant, that algorithms misclassify traffic in 5% to 6% of cases, potentially disrupting unrelated business services when hosting subnets are blocked.
Written from coverage as of
Images
- TechRadar
- TechRadar