Elon Musk’s latest changes for X are driving more users away – not exactly a surprise, granted – and many of them are flocking to rival social media outlet Bluesky. So many made the switch, in fact, it led to Bluesky briefly going down due to the volume of incoming new users.
The central move initiated by X that made the headlines for driving migration away from Musk’s platform is a change to the way the ‘Block’ button works. This was actually announced back in September, but is officially being implemented now (well, it’ll be in place ‘soon’ we’re told).
It means that going forward, X users who you have blocked will still be able to view your (public) posts – though they won’t be able to engage with them in any way (from replies to liking and so forth).
This is problematic for obvious reasons, in terms of enabling stalkers and trolls who will still be able to view the posts of an account that has blocked them, when previously this wasn’t the case. In the past, blocking meant that the blocked user couldn’t see any posts (or anything at all, save for a message telling them that they’ve been blocked), but soon, this will change.
Bluesky posted to say it had in excess of 100,000 new users inside 12 hours following the announcement by X, after the rival network highlighted the fact that its block function stops those who are blocked from viewing any posts.
In an update, Bluesky noted that it has now gained half a million new users in the past day.
There’s another reason that some folks are rapidly exiting from X stage left (and right, and indeed center, clambering over the audience, it would seem), and that’s a change to X’s privacy policy.
As TechCrunch reports, the new policy includes an update that allows third-party collaborators to use content on X to train their AI models – unless the user opts out. This is a notable extension of the reach of AI training on X, which has so far only been used to train Musk’s own Grok AI (unless users opt out, again).
Lol, that’s probably 30% of real people still using it
What a garbage dataset to train off. The majority of everything there is all bots and AI anyway lol
bots training bots. maybe this will (hopefully) corrupt the AI’s data. its kind of like copying off a copy again and again.
Sound like human
50und 11k3 hum4n
50()|V[) 11X3 |-|()|/|4|V
$[]|||||) |!|<[- #|_|///-||
Don’t forget to yell into the data PENIS PENIS PENIS PENIS PENIS PENIS!!!
Man I think it’s already done, and been done for a long time, without AI. So much of who we are now is based on and informed by what we’ve read and watched online. It’s our been our dominant frame of reference, language and value system for quite a while. We are the AI.
And that’s on top of all the Brazil users who fled to BlueSky when he refused to comply with a court order (and pay a fine) so they blocked Twitter for a few days. I’m not sure how many went back after he paid the fine but BlueSky was fairly popular in Brazil even during the closed beta so I’m sure a ton stuck around.
has so far only been used to train Musk’s own Grok AI
The Grok models are a laughing stock in the LLM space. They aren’t good over APIs, and they’re even less useful after Elon “open sources” them far later. Qwen 72B, and heck, Qwen 32B is already better than Grok 2, which is probably hundreds of billions of parameters. Qwen is runnable locally right now, Apache 2.0, and released day one. Grok 1 is… well, I dunno, no one has even bothered to try hosting it for anything.
I dunno what Twitter is doing with all those H100s Elon hoarded, but it seems like a big waste so far. Its certainly nothing to help the open source/self hosting space or to “decensor” and “democratize” LLMs like Elon fans seem to think.
No one leaves Twitter. They always come back. People are so addicted.