Identifying and Tracking People By Their Phone App Fingerprint Across the Globe
hubie writes:
No one here will be surprised by the size and reach of the data broker industry and the desire to identify and track individual users. From a data standpoint, the majority of human interaction now happens on mobile devices, and these devices end up collecting and storing a very large amount of our information. US privacy laws allow sharing and selling of anonymized data, which are those that do not contain Personally Identifying Information (PII), and the much ballyhooed GDPR imposes strict inform and consent requirements on collecting and using personal information. To comply with these laws many entities simply throw away all the personal information and keep the rest of the data, which sounds like a great thing until one realizes that it turns out to be relatively easy to deanonymize this data using otherwise innocuous data on the mobile device anyway.
The reason that so many apps report to third parties what other apps are installed on a phone is because these other apps create a remarkably unique fingerprint of each person. The aggregate of the installed apps and their relative usage turn out to be very unique for over 99% of people. So ad blockers can block cookies, and users can reset things like the Android Advertisement ID, but users can't reset or fake their app usage stats.
Some researchers looked at a data set of millions of people spanning 12 months and 33 countries and found that 91% of the people could be identified by looking at just the usage of four apps. They also looked at seasonal and cultural differences for identifying users. They found that people have more unique app fingerprints during summer months making them easier to identify then. They also saw significant variations in uniqueness across countries and found that American users are the easiest to identify while Finns have the least unique app fingerprints.
Sekara, V., Alessandretti, L., Mones, E. et al. Temporal and cultural limits of privacy in smartphone app usage [open]. Sci Rep 11, 3861 (2021).
DOI: 10.1038/s41598-021-82294-1
Read more of this story at SoylentNews.