Highlighting the recent report of users and admins being unable to delete images, and how Trust & Safety tooling is currently lacking.

  • maynarkh@feddit.nl
    link
    fedilink
    English
    arrow-up
    29
    ·
    8 months ago

    You are responsible for data collected by your own instance. If a deletion request comes through, you are responsible for deleting it from your account, and forwarding the deletion request and responses to other instance you federate with. You are in the clear as long as you don’t keep data you legally can’t, and have sufficiently informed other instances of your obligations.

    • RubberDuck@lemmy.world
      link
      fedilink
      English
      arrow-up
      6
      ·
      8 months ago

      No, if you collected the data and shared it with others, simply informing the others is not enough. This is why the platform needs tools for admins to comply.

      A proper method, that allows the users to nume their account could already be enough.

      • maynarkh@feddit.nl
        link
        fedilink
        English
        arrow-up
        4
        ·
        8 months ago

        What I mean by informing others is that you have to explicitly forward the deletion request. Not much else you can do I think.

        • RubberDuck@lemmy.world
          link
          fedilink
          English
          arrow-up
          2
          ·
          8 months ago

          I get that, but this is where it gets tricky. As “there is nothing we can do” was the number one reason used under the law predating the GDPR. So in the GDPR there is a stipulation that you stay responsible or share responsibility with the other party If you share the data. Because large companies used this to send data through clearing houses allowing them to hash their hands.

          GDPR is really the cranky brother of its predecessors, because there was so much fuckery going on.

          And while I doubt Admins will be a prime target for privacy watchdogs, it is good that they also have to think about the privacy of their users. Since privacy is a basic human right.

          • maynarkh@feddit.nl
            link
            fedilink
            English
            arrow-up
            1
            ·
            8 months ago

            Oh, that’s actually neat. But at the same time, that means every instance owner is responsible for the whole of the Fediverse.

            I can imagine that would mean non-compliant instances will get defederated at some point? Or ActivityPub will get some compliance features? It’s not like the EU is unaware of the Fediverse, they are the main monetary supporters behind Lemmy.

            • RubberDuck@lemmy.world
              link
              fedilink
              English
              arrow-up
              1
              ·
              8 months ago

              I have no clue how jurisprudence would turn out. But keep in mind, this is not about the posts people make. The framework just needs to collect/store as little information as possible that can be considered PII. And it should have a way to remove it.

              If Deleting your account results in the PII actually being removed (username, ip address, other profile info, whatever data is stored under the hood) and these removals actually get federated… there should not be an issue.

              Then admins maybe have to do something if people start posting PII as messages, but that would probably be doxing and up for removal anyway.

              So mainly the issus boil down to:

              • is there a way for people to scrub their account
              • does the scrubbing remove all the data
              • is the platform clear about what data is being collected and is all collected data actually needed
              • maynarkh@feddit.nl
                link
                fedilink
                English
                arrow-up
                1
                ·
                8 months ago

                The issue I see is that if my instance is on the hook for the fediverse at large, and I operate on an allowlist basis, malicious actors can scrape PII and ignore the GDPR, and that would make me the one on the hook for that, isn’t that right?

                • RubberDuck@lemmy.world
                  link
                  fedilink
                  English
                  arrow-up
                  1
                  ·
                  8 months ago

                  There is plenty of jurisprudence and clarity needed, so… maybe. Hence the importance for the framework itself to be as GDPR compliant as possible and not store PII if not nessecary and remove it once no longer nessecary. (Storing someone’s IP for login, and post validation, bans etc should be limited to the period that makes sense, not infinitely.)

                  And in your example, the ‘malicious’ part of the 3rd party probably makes it different. Maybe then it is a dataleak.