Mouse movements, typing speed, and typing style can serve as unique identifiers, much like fingerprints. As AI technology advances, it may become increasingly effective at recognizing these patterns, potentially compromising individuals’ anonymity. Are there any measures available to protect against this?

  • SwizzleStick@lemmy.zip
    link
    fedilink
    English
    arrow-up
    5
    ·
    5 days ago
    1. Don’t use shit that monitors in such a way.
    2. Where you can’t avoid it, do your work in a separate, trusted program and then copy paste it in.
    3. Where you can’t paste, I’m sure you could whip up an ahk script to ‘type’ for you from the clipboard.
    4. If you’re at this step, they’ve already won.
    • Palacegalleryratio [he/him]@hexbear.net
      link
      fedilink
      English
      arrow-up
      1
      ·
      edit-2
      5 days ago
      1. if they work reliably these tracking techniques will become pervasive across the internet.

      2. this is one of those things where the more you anonymise your data the more identifiable it is compared to everyone else’s non anonymous data.

      3. see above

      4. yeah…

      • SwizzleStick@lemmy.zip
        link
        fedilink
        English
        arrow-up
        1
        ·
        5 days ago

        The majority of people will sacrifice security for convenience, I guess we all know that’s a given.

        For 2 at least if enough users behave ‘adversely’ in the same way, they become less distinguishable from one another within that subset.

        Interesting to see how such an arms race pans out.

        • Palacegalleryratio [he/him]@hexbear.net
          link
          fedilink
          English
          arrow-up
          1
          ·
          5 days ago

          Yeah but that is still pretty identifiable, the same way fingerprinting works pretty well for finding a user with a specific screen resolution, os, fonts, etc that already makes a specifically small subset of users that you can start to find patterns and identify individuals. The tiny subset of users who might be tempted to go about anonymising inputs is already thinning the field huuugely

          • SwizzleStick@lemmy.zip
            link
            fedilink
            English
            arrow-up
            1
            ·
            5 days ago

            I’ve begrudgingly accepted that the greater war in this regard is long since lost, but I’m happy to win the small battles today. Unfettered internet access. Ad free media. VPNs, adblockers etc etc

            It helps that I’m a fairly dull person to begin with. I’ll probably be dust by the time it gets too bad to tolerate.

  • who@feddit.org
    link
    fedilink
    English
    arrow-up
    4
    ·
    edit-2
    5 days ago

    Are there any measures available to protect against this?

    • Disable scripts in your web browser, except for important sites that depend on javascript.
    • If any of those important sites are not trustworthy, write text in a local editor and copy/paste it into the site where needed. Do not type directly into text fields on the site.
    • Avoid other sites that require scripts.
    • Ask and/or pressure site owners to make them work without scripts where possible.

    Keep in mind that however you modify your text entry habits on untrusted sites, if scripts are allowed, your modified behavior can also be fingerprinted.

  • LittleBobbyTables@lemmy.sdf.org
    link
    fedilink
    English
    arrow-up
    1
    ·
    edit-2
    4 days ago

    There is typing speed ‘obfuscators’ available, such as kloak, and it’s also available in Whonix: https://github.com/Whonix/kloak

    It also obfuscates mouse movements (to a degree). But the Whonix docs say it’s not as effective as the typing speed obfuscation kloak provides. There’s some more context in their docs if you’re interested: https://www.whonix.org/wiki/Keystroke_and_Mouse_Deanonymization

    As for typing style, I assume you mean stylometry, Whonix also has a page on that here: https://www.whonix.org/wiki/Stylometry

    For that, the best known mitigation just seems to be to intentionally change up how you write in certain ways, or to “imitate the style of other known authors” (whatever that means). The docs say that “automated methods like machine translation services” aren’t a viable mitigation, but I’m not sure if that includes something like using AI to rewrite your sentences.