Human-seeded attacks and exploiting hot-spots in graphical passwords
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
Although motivated by both usability and security concerns, the existing literature on click-based graphical password schemes using a single background image (e.g., PassPoints) has focused largely on usability. We examine the security of such schemes, including the impact of different background images, and strategies for guessing user passwords. We report on both short- and long-term user studies: one lab-controlled, involving 43 users and 17 diverse images, and the other a field test of 223 user accounts. We provide empirical evidence that popular points (hot-spots) do exist for many images, and explore two different types of attack to exploit this hotspotting: (1) a “human-seeded ” attack based on harvesting click-points from a small set of users, and (2) an entirely automated attack based on image processing techniques. Our most effective attacks are generated by harvesting password data from a small set of users to attack other targets. These attacks can guess 36 % of user passwords within 2 31 guesses (or 12 % within 2 16 guesses) in one instance, and 20 % within 2 33 guesses (or 10% within 2 18 guesses) in a second instance. We perform an image-processing attack by implementing and adapting a bottom-up model of visual attention, resulting in a purely automated tool that can guess up to 30 % of user passwords in 2 35 guesses for some instances, but under 3 % on others. Our results suggest that these graphical password schemes appear to be at least as susceptible to offline attack as the traditional text passwords they were proposed to replace. 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