Re:HTML5 Canvas Fingerprinter
by mwexler on Wednesday January 14, 2015 @03:03PM
(#8888656)
I think some of the comments re: browser agent are missing the point. I think the poster was not focused on whether it can identify the system or browser type, but instead as a technique to provide a potentially unique (or somewhat unique) fingerprint that a user can't eliminate easily. So, a site could use this code (along with other techniques) to identify your browser and even if you delete a set cookie, detect that it's the same browser.
They may discover, btw, that it's not as unique as they think it is, but combined with other techniques, can be a tracking attribute that is difficult to disable.
It doesn't seem resilient to change in the underlying system from my cursory view (if I change my graphics card, resolution, or color depth, does this id change?) but like https://panopticlick.eff.org/, it's useful and a good cautionary tale to see the variety of ways your browser can be uniquely distinguished, with and without the use of cookies.
BTW, the icons on the left side of the browserleaks page are actually a menu and offer other "leaks" of info from the browser: some well known, others more subtle.
Re:HTML5 Canvas Fingerprinter
by falcolas on Wednesday January 14, 2015 @02:11PM
(#8888310)
Note that this is browser fingerprinting, not necessarily the user of the browser. Useful to detect spoofed User-Agent strings, not sure to what end though.
Re:HTML5 Canvas Fingerprinter
by davesque on Wednesday January 14, 2015 @03:59PM
(#8888998)
The fingerprint is really on the browser, graphics driver (os-specific), and graphics hardware. Other sources of information could be used to increase the probability that the fingerprint is unique to the user.
Re:HTML5 Canvas Fingerprinter
by aselzer on Wednesday January 14, 2015 @02:58PM
(#8888624)
It doesn't need it, it stores the User Agent beside the unique fingerprint to tell people with the same fingerprint what browsers they (probably) have.
If it really doesn't work without UA, then it's just the website not expecting someone to have disabled it.
They may discover, btw, that it's not as unique as they think it is, but combined with other techniques, can be a tracking attribute that is difficult to disable.
It doesn't seem resilient to change in the underlying system from my cursory view (if I change my graphics card, resolution, or color depth, does this id change?) but like https://panopticlick.eff.org/, it's useful and a good cautionary tale to see the variety of ways your browser can be uniquely distinguished, with and without the use of cookies.
BTW, the icons on the left side of the browserleaks page are actually a menu and offer other "leaks" of info from the browser: some well known, others more subtle.
http://en.wikipedia.org/wiki/Canvas_fingerprinting
Victory! HA!
[enables javascript]
> General conclusion: Your system fingerprint appears to be unique, yet we don't collect signatures here, just check.
Interesting. That is really not the answer I'd like to see. Is there a "blend in more" plugin?
There's a Chrome extension called Chameleon ( https://github.com/ghostwords/chameleon ) that attempts to cut down on fingerprinting.
(Aurora/Developer Edition on Arch Linux)
I understand that they can figure out that I'm running some variation of FF, but .. why would it think this is Windows?
I am also using FF Developer Edition on Arch Linux.
It says Firefox on Linux to me.
Signature B15C225D
would be me, for what it's worth.