I'm thinking about most of my communications on Telegram and it's kind of hilarious to imagine it suddenly starting to reply to people on Xitter with unsolicited horny furry roleplay instead of unsolicited white supremacist rants.

Not so hilarious that it doesn't make me want to consider trying to convince all my circles to move to a community-run Matrix server or something though.

Looks like we're at the point where it's both convenient and useful to destroy and recreate entirely new digital identities every few months.

The only cheap way out for users is to generate noise. Clog up their systems with useless data.

Yeah, sounds like they finally accepted that tweets don't make LLM smarter, only less biased, and so they're doubling down to feed more tweet-like data to make it smart and biased.

No I am sure the fine and morally upstanding leadership of xAI are paying $300M just for the fun of it.

Why wouldn't you automatically assume this is the case?

Because it's expensive and telegram already is bot traffic to a large portion (channels, bots, ...) that would be completely useless for training.

I wouldn’t trust X or Elon Musk, and $300M is the kind of money that you can ask for a lot of things from the recipient.

Yes it's some rich Russian dude trading with Elmo.

FTFY.

Yes, it's some ̵r̵i̵c̵h̵ ̵R̵u̵s̵s̵i̵a̵n̵ ̵d̵u̵d̵e̵ ̵t̵r̵a̵d̵i̵n̵g̵ ̵w̵i̵t̵h̵ Russian president calling Elmo.

Did you have an expectation of privacy using Telegram? I think that's the real issue here. If you were paying for Telegram, then I would say yes, those conversation should not be handed over to xAI. But, if you were ostensibly not paying for it, then I think it should be assumed that use of the service is considered consent to take that and use it as they please.

So yeah, it's almost certain that everything you ever put into Telegram is in Grok.

That’s not at all how consent works