There are generally 2 camps when it comes to browser fingeprinting, as far as I know
- reduce data as much as possible: ResistFingerprinting (RFP), TorBrowser, Librewolf
- randomize identifiers: Brave Browser
Because of extension support on mobile, and because or future-proof mv2 extensions, and because it is awesome, I stick with Firefox.
Ironfox on GrapheneOS, Firefox with custom policy.json on Linux.
The policy contains most of the arkenfox tweaks but not all. Policies can’t actually change some settings so you need an autoconfig (globally) of a user.js per user per profile.
Generally, it is hardened as much as feasible. As many identifiers as possible are reduced and randomized, but what to do about the leftovers?
Aliexpress & Audio context fingerprinting
Turns out this wasn’t new at all and there was a firefox extension protecting against this for quite some time! The developer IlGur has a bunch of other extensions that report fake values.
- AudioContext
- Fonts
- WebGL
- Canvas
I now use these together with a (changing) VPN, a browser that deletes cookies and other data between sessions for all but allowlisted sites, UBlockOrigin in expert mode blocking as much javascript as possible from even loading, on a hardened browser.
Still, this is only good if the extensions actually work and report actually random data, and if enough people use them so you can blend in.
Do you know of studies about this? How random is a canvas fingerprint for example, would a randomized one stand out?