Habr•September 10, 2026•🇷🇺Translated from Russian

Amnezia VPN Survives Coordinated Russian Censorship Campaign Targeting AmneziaWG Protocol Fingerprints

Amnezia VPN has released a comprehensive account of the intense blocking campaign Russian authorities waged against its services this summer, detailing how the company restored stable operation of Amnezia Free and Amnezia Premium after repeated waves of sophisticated censorship.

Attack Overview

The campaign was not a conventional protocol block but a coordinated infrastructure attack. Operators observed targeted reconnaissance of the API, service behavior, and infrastructure logic, accompanied by DDoS attempts and direct attacks on the API. While the API itself resisted compromise thanks to prior audits, individual components were successfully flooded. The company also faced DDoS against its website; because it uses Amazon CloudFront, the 100k–500k RPS peaks were mitigated through rapid rate-limit activation by Amazon.

Roskomnadzor changed its blocking strategy. Instead of blocking protocols outright, censors now fingerprint traffic patterns, map servers receiving matching traffic, and automatically blacklist addresses and entire subnets. Blocks are rolled out sequentially rather than simultaneously, allowing the service to recover from one wave before the next set of rules is deployed.

Technical Mitigations Implemented

Amnezia closed several protocol fingerprints: zero-length UDP packets in certain clients, fixed-size keepalive packets, handshake timing artifacts, nonce zero bytes, and related infrastructure detection vectors. After multiple client updates, mass blocking of newly deployed servers largely ceased. Connections from outdated Amnezia Premium clients are now disabled. Self-hosted AmneziaWG installations were mostly unaffected by protocol-level rules, though some users still encountered subnet blocks.

Timeline of the Incident

The first wave began on the night of 20 May with numerous servers blocked alongside DDoS and apparent replay or scanning activity against AmneziaWG. Subsequent waves on 1 June, 7 July, 10 July, and 25 July combined IP blacklisting, subnet monitoring, API scanning, phishing attempts against staff, and automated detection of fresh servers within hours of deployment. The active phase lasted roughly six weeks.

Engineers discovered that AmneziaWG 1.0 and 1.5 versions were identifiable by fixed packet sequences following a predictable initial handshake. This prompted an accelerated rollout of AmneziaWG 2.0, which dropped support for Windows 7, Android 8, and certain router configurations. Additional fixes addressed keepalive packets and zero-length UDP packets on specific platforms.

Future Defenses and Infrastructure Changes

Work on AmneziaWG 3.0 is underway to eliminate remaining structural features such as packet-size sequences, inter-packet timing, post-handshake behavior, and repeatable combinations that allow cumulative scoring. The company is also doubling server capacity for its VLESS infrastructure to provide resilient fallback connectivity that survives both protocol and IP-based blocking.

Related articles

Habr•Privacy & Surveillance

Dynamic Sites on Yandex Cloud Storage Enable Communication Under White-List Restrictions

A developer facing mobile internet white lists explored ways to exchange encrypted messages without relying on blocked infrastructure. Initial attempts with a national messenger bot failed due to poor API documentation and account hijacking. The team then modeled the problem as three neighboring plots where Alice and Bob route messages through Eve while a slow postman handles delivery. Yandex Disk was tested as a message store accessible via direct GET and POST requests, though it required accounts and suffered from 15-25 second latency. The final approach used Yandex Cloud object storage, whose domains remained whitelisted, allowing HTML files to load and function even when only government services and select platforms were reachable.

AntiMalware•Privacy & Surveillance

Rawline Aggregates VK, MAX, DNS and Other Tunnels into One Multi-Platform VPN Client

Rawline is a new multi-platform client that combines various VPN protocols and non-standard transports into a single application for Android, Windows, Linux and macOS. The project is built on Throne using sing-box and Xray network cores and supports both conventional protocols and unusual transports such as VK and MAX calls, WebRTC, DNS and S3. Users can import profiles via links or QR codes, test latency, chain connections and apply per-app or per-domain routing rules. The developer paid special attention to security issues, removing unnecessary local proxy endpoints, protecting internal connections with random passwords and adding traffic ownership verification that is enabled by default on Android 10 and newer. Additional fixes were implemented after discovering that some transport bridges could be hijacked by external applications sending the first UDP packet. Source code and builds are available on GitHub.

Habr•Privacy & Surveillance

Russia Freezes Entire Hosting Networks to Block VPN Traffic, Investigation Finds

An independent test of 435 endpoints across 154 hosting providers revealed that Russian internet providers are throttling connections to entire autonomous systems rather than individual IP addresses. Networks belonging to Hetzner, OVH, DigitalOcean, M247 and HostPapa are frozen after roughly 25 TCP packets, causing downloads to stall at around 16 KB while UDP-based WireGuard tunnels remain unaffected. The blocking is applied uniformly by multiple Russian operators including Beeline, MTS and home broadband providers, indicating a centrally distributed rule from Roskomnadzor via the TSPU filtering system. Smaller providers that rent IP space from the affected autonomous systems are also impacted, regardless of their own brand or location. Cloudflare-hosted resources currently pass the test, but the list of blocked networks has remained stable since mid-2025. The researcher published an open-source script that allows anyone to reproduce the measurements in under a minute.

Habr•Privacy & Surveillance

macOS Builds Undocumented Personal Knowledge Graph Since Version 13, Inferring Family Ties via On-Device ML

Since macOS 13 in 2022, two daemons named knowledgeconstructiond and intelligenceplatformd have been constructing a comprehensive personal knowledge graph on every Mac. The system aggregates data from Contacts, Calendar, Photos, Messages, Mail, Family Sharing, Health emergency contacts, Screen Time, and behavioral signals to map users, their relationships, locations, and activities. Results are stored in ~/Library/IntelligencePlatform as SQLite databases, including graph.db containing 181 entity classes and 278 predicates. Two CoreML models, EntityTagging_Family.mlmodelc and EntityTagging_FamilyAndFriends.mlmodelc, use 144 behavioral and photographic features to classify family and friend relationships with daily updates in entityTagging.db. An embedded LLM and embedding models further label unnamed faces and infer relationships from message text when Apple Intelligence is enabled. The graph remains queryable locally and is designed to feed personalized AI features while operating entirely on-device under Full Disk Access protections.