Researchers at Indiana University and Penn State University have announced a simple and fast method capable of detecting much of the malware used in zero-day attacks.
Their method is presented in the paper Finding Unknown Malice in 10 Seconds: Mass Vetting for New Threats at the Google-Play Scale[1]. I have not yet read it carefully myself, but from a quick look, it appears to detect malware by making use of the following observations:
- Most Android malware is repackaged existing malware => this can be identified by examining Java methods.
- It is suspicious when unrelated apps use the same methods, except for methods from well-known libraries.
In other words, to put it very roughly, the process seems to be:
- First, prepare a list of methods from well-known libraries (A).
- Extract the methods from each apk file in the Google Play Store (bi).
- For each i, calculate ci=bi\A.
- For every i and j, check whether ci ∩cj≠∅.
It seems that this is how the check is performed. If ci ∩cj≠∅, the apps are considered suspicious.
This approach seems likely to produce a great many false positives, but apparently it does not produce that many in practice. In addition, code used for zero-day attacks tends to spread through multiple apps, so this method is reportedly able to detect such code as well.
Under the Google Play Store’s current review system, even when a piece of malware is excluded, only the developer is excluded. As a result, the same code is sometimes submitted by another applicant and passes review. Incorporating a mechanism like this into the review process therefore seems likely to be extremely useful.
I have not had enough time lately to write about this in detail, but if you are interested, why not read the original paper?
[1] Chen, K., et al: Finding Unknown Malice in 10 Seconds: Mass Vetting for New Threats at the Google-Play Scale, Indiana University (2015/8/31) http://www.informatics.indiana.edu/xw7/papers/vetfast.pdf
Related posts

WPAD/PAC Attacks Against OAuth and Countermeasures
The WPAD/PAC Attack presented on August 3 at Black Hat 2016, which can read HTTPS URLs, is an attack that makes you think, “Ah, I see.”…
Deepfake Damage Reaching ¥6 trillion? The Impact of Generative AI on Identity and Countermeasures
It appears that the economic damage caused by deepfakes is becoming staggering. https://www.youtube.com/watch?v=GLNJKtM3-Io According to the SBBIT article “The ‘Improved Accuracy’ of Deepfakes Is Extremely Alarming:…

IETF 123: OAuth WG Session 1 Summary (Japan Time, Day 24)
Overview On July 24 Japan time, OAuth WG Session 123 at IETF 1 Madrid was held. The previously announced agenda was as follows, but a session…
