I don’t have Threads or follow anyone from there but sounds a bit… complicated?

https://techcrunch.com/2025/06/17/threads-expands-open-social-web-integrations-with-fediverse-feed-user-profile-search/

I guess they’d argue that none of those pesky little data protection laws apply to competent authorities like the police and they’re probably justifying it with the criminal prosecution clause.

Great.

VC financed social media at the start of its enshittification journey. Started interesting with open protocol but seems it has no ambitions of being open.

Deleted my account 6 months ago and haven’t missed it.

LLMs are a tool. They lack fidelity and frequently generate wrong results but as long as you (human subject matter expert) go through the results they are extremely useful in analysing and summarising large datasets.

But that’s all they are. They don’t create anything new, you can’t use them for learning anything and they certainly don’t ”think”, they just produce nice looking sentences.

Generally, the energy needs and environmental harm they do make their usecases very narrow. But there are some.

As a general tool that pretends to be ”intelligent”, they are completely useless because they’re nothing of the kind.