Mark Newton (@NewtonMark@eigenmagic.net)
mastodon.sdf.org
Attached: 1 image Slack is now using all content, including DMs, to train LLMs. They offer a thing they're calling an "opt-out." The opt-out (a) is only available to companies who are slack customers, not end users, and (b) doesn't actually opt-out. When a company account holder tries to opt-out, Slack says their data will still be used to train LLMs, but the results won't be shared with other companies. LOL no. That's not an opt-out. The way to opt-out is to stop using Slack. https://slack.com/intl/en-gb/trust/data-management/privacy-principles

cross-posted from: https://lemmy.ml/post/15741608

They offer a thing they’re calling an “opt-out.”

The opt-out (a) is only available to companies who are slack customers, not end users, and (b) doesn’t actually opt-out.

When a company account holder tries to opt-out, Slack says their data will still be used to train LLMs, but the results won’t be shared with other companies.

LOL no. That’s not an opt-out. The way to opt-out is to stop using Slack.

https://slack.com/intl/en-gb/trust/data-management/privacy-principles

  • Azzu@lemm.eeEnglish
    0·
    7 months ago

    I’m sorry, can everyone not read the actual link? They specifically say they are not training generative AIs, i.e. LLMs.

    They are using the data to train non-generative AIs for stuff like emoji and channel suggestions. I.e. “you use this emoji a lot, so it’s displayed first” or “people that are in these channels of yours also join these other channels you’re not in yet”.

    This is class A misinformation being spread here, good job. It’s unbelievable that I’m the first one to actually verify the truth of this post, because I wanted to share it further and that’s what you do before.

    • gila@lemm.eeEnglish
      0·
      7 months ago

      It also explicitly states in the posted screengrab that the opting-out user’s workspace won’t contribute to the underlying models. How would that be separate from using info on their workspace as training data for any kind of model? My interpretation of that is the data would be used to inference on the models, not train them.