Effective attacks and provable defenses for website fingerprinting
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Website fingerprinting attacks allow a local, passive eavesdropper to identify a user’s web activity by lever-aging packet sequence information. These attacks break the privacy expected by users of privacy technologies, including low-latency anonymity networks such as Tor. In this paper, we show a new attack that achieves sig-nificantly higher accuracy than previous attacks in the same field, further highlighting website fingerprinting as a genuine threat to web privacy. We test our attack under a large open-world experimental setting, where the client can visit pages that the attacker is not aware of. We found that our new attack is much more accurate than previous attempts, especially for an attacker monitoring a set of sites with low base incidence rate. We can correctly de-termine which of 100 monitored web pages a client is visiting (out of a significantly larger universe) at an 85% true positive rate with a false positive rate of 0.6%, com-pared to the best of 83 % true positive rate with a false positive rate of 6 % in previous work. To defend against such attacks, we need provably ef-fective defenses. We show how simulatable, determinis-tic defenses can be provably private, and we show that bandwidth overhead optimality can be achieved for these defenses by using a supersequence over anonymity sets of packet sequences. We design a new defense by ap-proximating this optimal strategy and demonstrate that this new defense is able to defeat any attack at a lower cost on bandwidth than the previous best. 1
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it