DeResistor's two-objective fitness function (balancing evasion success and detection probability) reduces flow-level detection rates from 96.27% → 45.06% against China's GFW, 99.50% → 34.93% against India, and 99.50% → 49.22% against Kazakhstan over 5 training generations, while in all cases preventing TRW from reaching an IP-block decision that would terminate training.

DeResistor-generated evasion strategies achieve an overall success rate of up to 98.61% against GFW (across vantage points in Qingdao, Shanghai, and Beijing) for the best strategy, and 100% in both India (Bangalore) and Kazakhstan (Oral) for the top-performing strategy, while standalone Geneva strategies tested in the same environment achieve comparable or slightly lower rates on some censors but are blocked at the IP level before training completes.

§6.3, Table 2 evaluation

Geneva packet-manipulation probing traffic exhibits distinctive features — corrupt data-offset fields, smaller packet sizes, overlapping TCP segments, TTL variance, and non-zero SYN packets — that allow simple ML classifiers (Decision Trees, Random Forests, Logistic Regression, SVM) to detect it with AUC > 0.99. A subsequent TRW-based IP-level detector can then block the source IP with high confidence after inspecting only 2 Geneva probing flows.

§4.1–§4.2 detection

GFW employs layered blocking for high-value targets: DNS poisoning for domains like google.com and wikipedia.org combined with null-routing of their hosting IPs, meaning packet-manipulation tools that operate at the TCP/HTTP layer (e.g., Geneva, DeResistor) cannot generate or test evasion strategies because no response is received to the initial SYN — the blocking occurs below the layer those tools target.

§6.1 detection

Interleaving a single normal benign flow (jump size J=1) after each detected probe prevents the TRW likelihood ratio from converging to the IP-block threshold across all 11 simulated censors and all three real-world censors tested; setting J>1 risks triggering a history-aware TRW reset that can paradoxically accelerate IP-level detection.

§5.2, §6.2 defense

Stage 1 of the detection pipeline uses a lightweight heuristic: restrict analysis to IP addresses in "VPS-dense ASNs," which censors already target for resource-intensive inspection of fully-encrypted traffic. This pre-filter dramatically reduces the search space before applying the more expensive dual-role behavioral analysis. The evaluation was conducted without Stages 1 and 3 due to dataset limitations, meaning the reported 23% recall and 0.18% FPR are conservative lower bounds on the full pipeline's performance.

2026-almutairi-server §2.1, §2.2, §3.4 fully-encrypted-detectdpiml-classifier

The host-profiling censor (passive traffic analysis: count connections per server, block those exceeding a threshold τ within a window w) blocks essentially all circumvention user traffic within 30 time steps for all classifier qualities tested (ρ_TP ∈ {0.9, 0.95, 0.99}), while causing far less collateral damage than zig-zag (never exceeding ~30% innocent server blocking). Snowflake resists this attack well: with w=3, τ=3, over 94.48% of users receive a proxy within 2 steps even with worst-classifier rules, and final unblocked server rates are 91.24–99.04%. The host profiling approach is feasible for passive censors who cannot enumerate the distribution channel.

2026-fares-game §5.2, Table 2, Table 3 traffic-shapeml-classifier

Balboa's synchronous leaf-content replacement adds non-negligible timing differences that allow censors to identify its activity with up to ~90% accuracy over different network conditions. The timing anomaly arises because Balboa performs data substitution directly at each data exchange, delaying the server's response while covert data is prepared.

2026-kamali-huma §I (Introduction), §II-A traffic-shapeml-classifier

Without chunk-based padding, an XGBoost classifier identifies the target website from covert data-chunk sizes with 91% accuracy (Tranco top-100). Chunking at 2 MB reduces accuracy to 12% at a 21.3% bandwidth overhead, while 16 MB chunks reduce accuracy to near random guessing at a 480.3% overhead. Chunks as small as 64 KB already reduce accuracy to 64%, demonstrating a monotonic fingerprinting–overhead tradeoff.

2026-kamali-huma §V-C, Figure 4 website-fingerprintml-classifier

Huma's deferred-reply / double-request receive (DRR) protocol reduces a traffic-fingerprinting XGBoost classifier's accuracy to at most 54% (near random guessing) across geographically distributed clients (San Francisco, Frankfurt, Bangalore). A Kolmogorov-Smirnov test on absolute page-load timing distributions yields D=0.03, p=0.98 for U.S. clients — substantially tighter than Waterfall of Liberty's D=0.11 at p=0.5 — confirming that Huma flows are statistically indistinguishable from benign HTTPS fetches.

2026-kamali-huma §V-D, Table II traffic-shapeml-classifier

When the researchers attempted to use Gemini 2.5 Flash as a third independent LLM judge via its API for evaluating moderation decisions, Gemini automatically blocked all judging attempts citing safety reasons. This occurred even though the research task (judging whether a response is more or less moderated) does not itself produce harmful content. The incident illustrates that LLM safety systems can over-block legitimate research use cases, and that different LLM providers have different thresholds— Claude Haiku 4.5 and GPT-4o completed all judging tasks without safety refusals.

2026-lipphardt-dual §3.3.3 ml-classifierkeyword-filtering