I have been testing Tor Browser and Mullvad Browser using fingerprint.com. I get unique persistent identifiers that are unique per machine and persist over rebooting sessions. Javascript was on during this test.

This could be very dangerous to people using Tor Browser and Mullvad Browser.

For example, if someone visits Rainbow Railroad, an organization for leaving repressive countries with hostile LGBT policies, and then watches a video about the organization on YouTube, and then also does something, like create a Discord Server, and use Tor Browser to get around geoblocking but link it to their personal phone number, then a hostile regime buying data from data brokers could possible determine that user is considering using rainbow railroad. Even if this exact example isn’t realistic or plausible (although governments do buy form data brokers), users should be aware that persistent identifiers in Tor Browser and Mullvad Browser allow for continuous tracking of a user using the same machine.

I posted this information on privacyguides forum and they deleted my account after, leading me to wonder if the forum is a giant honeypot that curates acceptable privacy discussions and unacceptable private discussions. I honestly wonder if they are infiltrated by the government. They repeatedly delete the posts of other people as well and the whole thing is starting to not sit well with me.

  • CorvusCornix@piefed.social
    link
    fedilink
    English
    arrow-up
    1
    ·
    3 days ago

    I know this post is a bit older at this point, but do you have any custom fonts installed? I could reliably reproduce my browser fingerprint as you described, even between Librewolf and Firefox with different configurations (including JavaScript disabled in Librewolf). A visit on one browser would increment the counter, and vice-versa.

    I had forgotten that, quite some time ago, I’d installed a font not packaged with my OS - I deleted it, made sure to clear out my site data, changed IPs, and finally got a shiny new fingerprint.

    Cross-referencing with EFF’s coveryourtracks site, I also noticed that one metric that almost never changes for me is my audiocontext fingerprint, although I can thwart that by disabling JavaScript, but maybe that’s contributing as well?