If only LLMs did that

Afaik the files exfiltrated were photos that the on device detection could not identify and were uploaded to verify server side. That would mean not all pictures are sent to the backend, and that corroborates why “only” 70k photos were stolen when discord has millions of users verified.
Of course you have to put your trust in a closed source system so best not to upload, but if true it’s still a far cry from openly lying about it. It’s probably explicitly stated in their ToS that they may upload the file if the verification fails client side.

Removed by mod

As long as you own the means of dishwashing, it’s okay?

The default banned instances list include lemmygrad and hexbear.
Pretty sane defaults, and the admins can change that after setting it up.
You can see it in app/cli.py inside the piefed git repo

I prefer to see dogshit human art than whatever soulless sinful shit generated by an AI.
Hell, most good memes are badly drawn dogshit, that’s part of the culture.

So I guess it’s backend problems? Why rewire the game client if the server needs rework…

I doubt the transistor on a GPU wafer break after 100k cycles, as they run at gigahertz frequencies, some cycle billions of times a second.

And you can take approximately 3 seconds to click on the kill switch if you so desperately need not to see an AI button somewhere.
Like I can understand (and I agree) the stance on AI in general, but this is just a knee jerk reaction. Your browsing experience is 99.9% unchanged even if there is a button somewhere…

Nobody wanted AI as a feature

This is false, you don’t speak for everybody and represent a small vocal minority of users.

It already is a broader ecosystem of trusted software. Have you looked at what is inside a modern browser? It’s not 1999 anymore, and tons of related stuff is embedded in a browser.

So if you never press the AI button, it’s never enabled. It is opt-in in the strictest semantic sense.
What you say here applies for things that run automatically, like the anonymous usage reports, which is opt out, not for things you activate yourself.

It’s a streamline I’m afraid

Tauri doesn’t automatically make apps perform good. Easy and common pitfalls still can make it go to a crawl just like electron.
Yaak is an example of a tauri app that performs horribly, and that can’t reach a satisfactory 30fps on modern hardware. The issue is within how tauri interacts with the js world and syncs state.

I saw a listing for 1340€ for 96gb. Prices are truly insane

Well, their customers are PC gamers so it makes sense to target them. Not all gamers build their PC themselves (I’d say most don’t as I have built most of my friends pc). Having a plug and play solution, without risk of hardware/driver/software issues, can be attractive to some. The market for these is not hardcore gamers, but couch gaming.
But if the hardware is more expensive than a pre-built, no one will buy it.
They said in interviews that some steam deck users primarily use it docked to a TV (like I do), and that showed them there is a market for such devices.

I would guess lower than 800 for the base model. The steam deck entry price is low and they probably will try to have the machine affordable. They know no PC gamer that can build their own PC will buy it if it’s not competitive.
Since they’re OEM integrating parts, I can guess around 550-650 for the base model but that will also depend on the US tariffs that week.

Watch the gamer nexus video if you want more info on the cooling. Iirc they use phase change TIM for the CPU and paste for the GPU.
They also talk about how they designed the case to avoid it being choked

Pretty sure this must have been in development before when the bug was discovered. You don’t make new features like this within a month.

You mean the security team that got pwned here?