Internal documents obtained by 404 Media show that Tumblr staff compiled users' data as part of a deal with Midjourney and OpenAI.

this could not be timed worse for Tumblr which is in huge hot water with its userbase already for its CEO breaking his sabbatical to ban a prominent trans user for allegedly threatening him (in a cartoonish manner), and then spending a week personally justifying it increasingly wildly across several platforms. the rumors had already been swirling that this would occur, but this just cements that they were correct

  • FaceDeer
    8•9 months ago

    I wouldn’t really trust that promise, frankly. I just checked their terms of service and it has the usual clause:

    You must own all rights, title, and interest, including all intellectual property rights, in and to, the User Content you make available on the Services. ASSC requires licenses from you for that User Content to operate the Services. By posting User Content on the Services, you grant ASSC a royalty-free, perpetual, irrevocable, non-exclusive, sublicensable, worldwide license to use, reproduce, distribute, perform, publicly display or prepare derivative works of your User Content.

    Which isn’t really surprising, it’s standard boilerplate for a reason. They don’t want to be caught in a situation where they can’t function legally any more. They say they won’t sell the company or your data, and they might even believe that right now, but who knows what the future might bring? They have the ability to do so if the circumstances arise.

    • fracture [he/him]
      2•9 months ago

      is there a way they could protect themselves (“still legally be able to function”) without that clause?

      • FaceDeer
        2•9 months ago

        You could ask a lawyer, I suppose. But the basic gist of this is “we don’t know what we might need to do with this data in the future, so we put ‘we can do anything with this data’ into the ToS so that we know that if the need arises we won’t find ourselves unable to do what we need to do with it.” Any website that doesn’t do this could find itself unable to implement new features or comply with new laws they didn’t think of when crafting the original ToS.

        At the very minimum a ToS needs to have some way to update and apply retroactively to old data, which ends up being “we can do anything with this data” with extra steps.

        • fracture [he/him]
          1•9 months ago

          thanks for sharing this information with us, i think it’s important to discuss this stuff on the fediverse

          i notice that beehaw doesn’t have a similar clause in its TOS, as far as i can tell. without the expectation of you answering this question, i’m wondering what the difference is between the two such that cohost has such a clause and beehaw doesn’t. maybe it’s because one is run by an individual and one is run by a small company?

          i did a search on cohost itself to see if anyone else talked about this and found this quite extensive thread: https://twitter.com/rahaeli/status/1588769277053739010

          so based on what you’ve said and what’s in that thread, i’m gonna update my post with some qualifications about cohost. thanks for piqing my interest in the TOS

          • FaceDeer
            2•9 months ago

            No problem. I’m not a lawyer myself, mind you, but I’ve encountered issues like these enough times over the years that I feel I’ve got a pretty good layman’s grasp. Plus I’ve actually read some of these ToSes and considered them from the perspective of the company running the site, which I suspect most people arguing about this stuff haven’t actually done.

            I wish the Fediverse sites running without rigorous ToSes well, of course, but I suspect failing to establish clear rights to use the content people post on them is likely to end up biting them in the long run. At least the bigger ones. Hobby-level websites get away with a lot because they don’t have significant money on the line.