SHAP analysis shows timing-based features — IEI standard deviation, mean click IEI, and time to first action — dominate agent identity classification under normal conditions, receiving substantially larger attributions than structural or action-type features. Agents are distinguishable primarily by their tempo: how long they pause before acting and how variable that pause is.

Strong classifiers can be trained from fewer than one third of available traces with gains diminishing rapidly beyond that threshold. At inference time, macro F1 rises sharply within the first 40% of observed actions across all four datasets, meaning model identity can be inferred while the agent is still actively navigating the page.

§6.2, Figure 6 detection

Passive JavaScript UI traces are sufficient to fingerprint the underlying LLM of a browser agent with up to 96% macro F1 across 14 frontier models, achieving roughly 10× random-chance accuracy. Even the weakest model pair (Qwen3.5-9B on 2WikiMultiHopQA) reaches 63.7% F1 against a ~7% random baseline for 14 classes.

§5.1, Figure 2 detection

In open-set fingerprinting (leave-one-agent-out protocol), the majority of models exceed AUROC 0.60 for unknown-agent detection, but closed-set and open-set performance are dissociated: Seed-2-lite achieves 96.1% closed-set F1 yet scores below-chance open-set AUROC (0.38–0.47 on three of four datasets), while GPT-5.4 achieves AUROC 0.84 open-set despite ranking third in closed-set F1.

§5.1, Figure 3 evaluation

Injecting uniformly sampled random delays between agent actions substantially degrades an unadapted XGBoost classifier, but a classifier retrained on delayed traces largely recovers performance across all four datasets. Under 5-second delay injection, the classifier shifts weight onto structural features (click-coordinate dispersion, structural key ratio, link-click ratio) that survive timing perturbation.

§6.1, Figure 5 evaluation

NATA (Non-invasive Active Traffic-correlation Analysis) injects low-frequency bandwidth waveforms (sinusoidal, square-wave, triangular) into Tor TCP connections at an upstream gateway without endpoint compromise, payload decryption, or Tor-browser modification. BM-Net, a selective state-space classifier trained on the exit-side observations, achieves a 99.65% binary detection F1 score distinguishing watermarked from natural traffic on a 20,000-trace real-world dataset.

2026-fan-activeflowmark-assessing-tor §IV, §VI.D, Table III flow-correlationtraffic-shapeml-classifier

BM-Net achieves a 99.65% binary detection F1 score for distinguishing bandwidth-watermarked Tor flows from natural traffic, outperforming all evaluated baselines (next best: TikTok at 75.96% F1). The accuracy gap stems from active perturbation imposing a deterministic low-frequency throughput constraint rather than relying on subtle natural metadata, making the detection task fundamentally easier than passive website fingerprinting.

2026-fan-activeflowmark-assessing-tor §VI-D, Table III–IV flow-correlationtraffic-shapeml-classifier

BM-Net achieves a 99.65% binary detection F1 score distinguishing watermarked from natural Tor flows, and a 97.5% macro-F1 score for fine-grained modulation classification across sinusoidal, square-wave, and triangular patterns. The fine-grained test set contains 201 held-out samples collected from ten clients across five geographic regions (Europe, North America, Australia, Southeast Asia, East Asia), with training traces including traffic collected under WTF-PAD and Walkie-Talkie defenses.

2026-fan-activeflowmark-assessing-tor §VI-D, Table III, Table V flow-correlationml-classifiertraffic-shape

BM-Net achieves a 97.5% macro-F1 score for fine-grained classification of three modulation geometries (sinusoidal, square-wave, triangular) from noisy exit-side Tor observations using only 201 labeled test samples collected across cross-continental Tor paths. Residual errors concentrate between natural traffic and square-wave modulation, as abrupt low-rate transitions are partially smoothed by Tor multiplexing and network jitter.

2026-fan-activeflowmark-assessing-tor §VI-D, Table V flow-correlationml-classifiertraffic-shape

Fine-grained modulation classification (natural vs. sinusoidal vs. square-wave vs. triangular) achieves 97.5% macro-F1 on a 201-sample held-out test set. Square-wave waveforms are the hardest class (F1 = 95.7%), while sinusoidal and triangular each reach 99.0% F1, because abrupt square-wave transitions are partially smoothed by Tor multiplexing and network dynamics.

2026-fan-activeflowmark-assessing-tor §VI.D.2, Table V flow-correlationtraffic-shapeml-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