I would like to send every person who uses user data to train an LLM to jail. Ethics jail, where they all must learn basic ethics of user consent and data privacy.
Latest exhibit: Slack is training a *global* model on conversations in your space. To opt out you need to email them. This is unconscionable and is extremely likely to leak private information.
Thank you.
This is getting so ridiculous. I guess I'm sending them an email.
What a hostile user experience.
Actually, I probably need to do some searching and reading. We use slack as a pretty basic channel based group chat. Mattermost is probably way more feature rich than we need. I'll need to look and see what simple self hosted things I could setup instead.
I've really not used much of anything other than slack, as far as tools similar to slack goes anyway. Unless you consider things like Skype and Teams (yuck and fucking ugh, respectively).
You've tried and didn't like mattermost? I've always thought it was pretty much just an open source clone of slack. What didn't you like about that?
The Enshittification continues apace.
@jamuraa This is going to be fun for everyone who wants to search private passwords.
Autocompleting passwords from Slack conversation is going to be big …
@jamuraa The sudden rise of "Descriptive image caption for users a who are visually impaired" (Mastodon is guilty too) Is virtue signaling and a dog whistle for "let use your images and captions to train AI models and use in ReCaptcha".
They rely on your honesty though, but I am not sure you can poison a dataset by using wrong captions on purpose.
If it was ever about the visually impaired, they could have an AI create the caption anyway.
Marie
Irenes (many)
SP⟁CED GO⟁T
relsqui
Oborosaur
Jay
creohn 💥
Violet Madder
ArneBab
Addison
CitroenSleutelAap
Kris Hardy
13 barn owls in a trenchcoat
Owen Mehegan
Kaspi