Backend
SST-Guard: Detecting and Characterizing Server-Side Google Analytics in the Wild
arXiv cs.CR0 views
arXiv:2604.27497v2 Announce Type: replace Abstract: As web browsers increasingly implement anti-tracking features, the web tracking ecosystem has been shifting from the client-side to the server-side. Instead of sending requests directly to the tracker's endpoint, server-side tracking (SST) sends tracking requests to publisher-controlled or intermediary endpoints that then forward the information to trackers from the server-side. As a result, client-side tracking protections are becoming fragile because direct client-to-tracker requests may no longer be observed. In this paper, we investigate the server-side Google Analytics (sGA), the most widely deployed third-party tracking service on the web today. We present SST-GUARD, a multi-modal browser-based approach for detecting sGA. The key insight behind SST-GUARD is that common sGA deployments change the standard Google Analytics endpoints, but still leave semantic artifacts of data collection by Google Analytics in the browser, including identifiers, event metadata, cookies, and JavaScript state. SST-GUARD detects these artifacts using value-templates, which use regular expressions to match semantic value patterns across multiple modalities: network requests, cookies, and the window object, providing robustness against obfuscation of one modality. We validate SST-GUARD on Tranco top-10k websites, detecting 4.0% (401) sGA domains with over 93.0% accuracy across three modalities, with network request classifier demonstrating the highest accuracy (99.8%). Additionally, SST-GUARD identifies atleast 43.2% more websites that integrate sGA as compared to prior approaches. Deploying SST-GUARD at scale, we detect sGA on 6,314 of Tranco top-150K websites. Our analysis shows that many sGA deployments use first-party subdomains, direct A/AAAA records, custom paths, or encoded payloads that circumvent existing defenses. We reported our findings to the authors of EasyPrivacy, who have integrated them.
Source:https://arxiv.org/abs/2604.27497PreviousTriosecuris: Formally Verified Protection Against Speculative Control-Flow HijackingNextCross-Modal Backdoor Attack on Connector-based Multimodal Large Language Models
Related posts
Fast and Memory-Light Evaluation of Density-Based Clusterings with the DISCO Index in R
0arXiv stat.CO
Norm-equation methods for structured group-penalized regression
0arXiv stat.CO
Best New Roblox Games to Play and Explore This Month
0FOSS Bytes

Karpenter v1.15: GPU-aware provisioning, smarter node repair, and more
0AWS Containers Blog
Free Orlando, Florida Real Estate Dataset: 5231 Listings in Excel (2026-11-03)
0dev.to Data
I Asked Apple's Search API for 20 VPN Apps and Got 18. Asking for 200 Gets You 189
0dev.to Data