Effective Real-Time Android Application Auditing
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
Mobile applications can access both sensitive personal data and the network, giving rise to threats of data leaks. App auditing is a fundamental program analysis task to reveal such leaks. Currently, static analysis is the de facto technique which exhaustively examines all data flows and pinpoints problematic ones. However, static analysis generates false alarms for being over-estimated and requires minutes or even hours to examine a real app. These shortcomings greatly limit the usability of automatic app auditing. To overcome these limitations, we design AppAudit that relies on the synergy of static and dynamic analysis to provide effective real-time app auditing. AppAudit embodies a novel dynamic analysis that can simulate the execution of part of the program and perform customized checks at each program state. AppAudit utilizes this to prune false positives of an efficient but over-estimating static analysis. Overall, AppAudit makes app auditing useful for app market operators, app developers and mobile end users, to reveal data leaks effectively and efficiently. We apply AppAudit to more than 1,000 known malware and 400 real apps from various markets. Overall, AppAudit reports comparative number of true data leaks and eliminates all false positives, while being 8.3x faster and using 90% less memory compared to existing approaches. AppAudit also uncovers 30 data leaks in real apps. Our further study reveals the common patterns behind these leaks: 1) most leaks are caused by 3rd-party advertising modules; 2) most data are leaked with simple unencrypted HTTP requests. We believe AppAudit serves as an effective tool to identify data-leaking apps and provides implications to design promising runtime techniques against data leaks.
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 | 0.000 | 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