FaceDeer

Honestly, I started laughing my head off when I saw that ragged flap still moving and the Starship still maintaining attitude control with it. That's the sort of battle damage I expect to see in a science fiction show, I wasn't expecting SpaceX to bring that sort of thing into the real world too.

There's also a simple toggle to turn Rewind off in the settings menu.

People are really going bonkers over Rewind, it's almost a sort of mass hysteria at this point. Yes, it appears to be a very insecure and risky feature at this point. So just turn it off. There's lots of features in any OS that you can set up in ways that will make your system insecure, this is just a particular one of those. Microsoft isn't going to force it to be enabled, the ensuing legal shitstorm would be epic. I doubt they'll roll it out to a large audience in its current state.

The thing that drives me nuts is that I really do value that baby they're carrying around. It is precious. But I don't want to give the Internet Archive money just to funnel into the pockets of their lawyers and settlement payments to big publishers due to these unrelated quixotic battles.

I was hoping that the IA would have learned a lesson from losing this court case, they should have settled as soon as they could. I'm sure the publishers don't want the bad publicity of "destroying" the Internet Archive, they just want them to stop blatantly violating their copyrights. But this appeal suggests that they haven't learned that lesson yet.

In an ideal world there'd either be some kind of leadership shakeup at the IA to get rid of whoever was behind this stunt, or some kind of alternative IA-like organization appears to pick up the archive before the IA goes broke and its collection ends up being sold off to the highest bidder. Or simply destroyed.

They don't need to do anything so drastic. They just need to stop doing things that blatantly provoke legal attacks like this. Their "Emergency Covid Library" was a foolish stunt that is endangering their primary objective of information preservation, they wouldn't have been sued if they'd just kept on carrying on as they were before.

It's not even a question of being "owned by corporations". Judges don't care about petitions. They're not politicians, their job is to adjudicate the law.

Except it's not a threat to the future of all libraries, it's a threat to the future of "libraries" that decide to completely ignore copyright and give out an unlimited number of copies of ebooks. Basically turning themselves into book-focused piracy sites.

I'm incredibly frustrated with Internet Archive for bringing this on themselves. It is not their mandate to fight copyright, that's something better left in the hands of activist organizations like the EFF. The Internet Archive's mandate is to archive the Internet, to store and preserve knowledge. Distributing it is secondary to that goal. And picking unnecessary fights with big publishing houses like this is directly contrary to that goal, since now the Internet Archive is in danger.

It's like they're carrying around a precious baby and they decided it was a good idea to start whacking a bear with a stick. Now the bear is eating their leg and they're screaming "oh my god help me, the bear is threatening this baby!" Well yeah, but you shouldn't have brought a baby with you when you went on a bear-whacking expedition. You should have known exactly what that bear was going to do.

Aside from it not really working, though.

Glaze attempts to "poison" AI training by using adversarial noise to trick AIs into perceiving it as something that it's not, so that when a description is generated for the image it'll be incorrect and the AI will be trained wrong. There are a couple of problems with this, though. The adversarial noise is tailored to specific image recognition AIs, so it's not future-proof. It also isn't going to have an impact on the AI unless a large portion of the training images are "poisoned", which isn't the case for typical training runs with billions of images. And it's relatively fragile against post-processing, such as rescaling the image, which is commonly done as an automatic part of preparing data for training. It also adds noticeable artefacts to the image, making it look a bit worse to the human eye as well.

There's a more recent algorithm called Nightshade, but I'm less familiar with its details since it got a lot less attention that Glaze and IIRC the authors tried keeping some of its details secret so that AI trainers couldn't develop countermeasures. There was a lot of debate over whether it even worked in the first place, since it's not easy to test something like this when there's little information about how it functions and training a model just to see if it breaks is expensive. Given that these algorithms have been available for a while now but image AIs keep getting better I think that shows that whatever the details it's not having the desired effect.

Part of the reason why Cara's probably facing such financial hurdles is that it's computationally expensive to apply these things. They were also automatically running "AI detectors" on images, which are expensive and unreliable. It's an inherently expensive site to run even if they were doing it efficiently.

IMO they would have been much better served just adding "No AI-generated images allowed" to their ToS and relying on their users to police themselves and each other. Though given the witch-hunts I've seen and the increasing quality of AI art itself I don't think that would really work for very long either.

I get the sense that a federated image hosting/sharing system would be counter to their goals, that being to lock away their art from AI trainers. An AI trainer could just federate with them and they'd be sending their images over on a silver platter.

Of course, any site that's visible to humans is also visible to AIs in training, so it's not really any worse than their current arrangement. But I don't think they want to hear that either.

And then that growth promptly blew its budget because it's using expensive cloud AI services from Vercel and it has no means of monetization whatsoever to bring money in.

People can do whatever they want, of course. But they have to pay for the resources they consume while doing that, and it seems Cara didn't really consider that aspect of this.

I've mainly been using open-weight models I can run locally to back them, so it'll last as long as personal computers do.

But I really do find them useful, so they are getting it right in at least some cases.

It's not all imaginary, real money is being paid for AI services which in turn is being spent on GPUs.