Mimicry Attacks on Smartphone Keystroke Authentication
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
Keystroke behaviour-based authentication employs the unique typing behaviour of users to authenticate them. Recent such proposals for virtual keyboards on smartphones employ diverse temporal, contact, and spatial features to achieve over 95% accuracy. Consequently, they have been suggested as a second line of defense with text-based password authentication. We show that a state-of-the-art keystroke behaviour-based authentication scheme is highly vulnerable against mimicry attacks. While previous research used training interfaces to attack physical keyboards, we show that this approach has limited effectiveness against virtual keyboards. This is mainly due to the large number of diverse features that the attacker needs to mimic for virtual keyboards. We address this challenge by developing an augmented reality-based app that resides on the attacker’s smartphone and leverages computer vision and keystroke data to provide real-time guidance during password entry on the victim’s phone. In addition, we propose an audiovisual attack in which the attacker overlays transparent film printed with spatial pointers on the victim’s device and uses audio cues to match the temporal behaviour of the victim. Both attacks require neither tampering or installing software on the victim’s device nor specialized hardware. We conduct experiments with 30 users to mount over 400 mimicry attacks. We show that our methods enable an attacker to mimic keystroke behaviour on virtual keyboards with little effort. We also demonstrate the extensibility of our augmented reality-based technique by successfully mounting mimicry attacks on a swiping behaviour-based continuous authentication system.
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 | 0.000 | 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 | 0.001 | 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