PatchLess

Invidious still seems to work for VODs provided the instance doesn’t get restricted. Livestreams have been broken for ages though.

I don’t really see the advantage here besides orchestration tools unless the top secret cloud machines can still share it’s resources with public cloud to recoup costs?

I have really mixed feelings about this. My stance is that I don’t you should need permission to train on somebody else’s work since that is far too restrictive on what people can do with the music (or anything else) they paid for. This assumes it was obtained fairly: buying the tracks of iTunes or similar and not torrenting them or dumping the library from a streaming service. Of course, this can change if a song it taken down from stores (you can’t buy it) or the price is so high that a normal person buying a small amount of songs could not afford them (say 50 USD a track). Same goes for non-commercial remixing and distribution. This is why I thinking judging these models and services on output is fairer: as long as you don’t reproduce the work you trained on I think that should be fine. Now this needs some exceptions: producing a summary, parody, heavily-changed version/sample (of these, I think this is the only one that is not protected already despite widespread use in music already).

So putting this all together: the AIs mentioned seem to have re-produced partial copies of some of their training data, but it required fairly tortured prompts (I think some even provided lyrics in the prompt to get there) to do so since there are protections in place to prevent 1:1 reproductions; in my experience Suno rejects requests that involve artist names and one of the examples puts spaces between the letters of “Mariah”. But the AIs did do it. I’m not sure what to do with this. There have been lawsuits over samples and melodies so this is at least even handed Human vs AI wise. I’ve seen some pretty egregious copies of melodies too outside remixed and bootlegs to so these protections aren’t useless. I don’t know if maybe more work can be done to essentially Content ID AI output first to try and reduce this in the future? That said, if you wanted to just avoid paying for a song there are much easier ways to do it than getting a commercial AI service to make a poor quality replica. The lawsuit has some merit in that the AI produced replicas it shouldn’t have, but much of this wreaks of the kind of overreach that drives people to torrents in the first place.

That’s not really fair on Discord. The article mentions they received an injunction to remove the content so they were forced to do this. Anybody in the same jurisdiction would have to do the same:

“Discord responds to and complies with all legal and valid Digital Millennium Copyright Act requests. In this instance, there was also a court ordered injunction for the takedown of these materials, and we took action in a manner consistent with the court order,” reads part of a statement from Discord director of product communications Kellyn Slone to The Verge.

It does have to do with being a walled platform though. You as the Discord server owner have zero control over whether or not you are taken down. If this was Lemmy or a Discourse server (to go with something a little closer to walled garden) that they ran, the hosting provider or a court would have to take them down. Even then the hosting provider wouldn’t be a huge deal since you could just restore backup to a new one Pirate Bay style. Hell, depending on whether or not the devs are anonymous (probably not if they used Discord), they could just move the server to a new jurisdiction that doesn’t care. The IW4 mod for MW2 2009 was forked and the moved to Tor when Activision came running for them so this isn’t even unprecedented.

Also holy shit: Club Penguin! This kind of thing has been around forever!

Good! You wanna automate away a human task, sure! But if your automation screws up you don’t get to hide behind it. You still chose to use the automation in the first place.

Hell, I’ve heard ISPs here work around the rep on the phone overpromising by literally having the rep transfer to an automated system that reads the agreement and then has the customer agree to that with an explicit note saying that everything said before is irrelevant, then once done, transfer back to the rep.

I don’t have a problem with training on copyrighted content provided 1) a person could access that content and use it as the basis of their own art and 2) the derived work would also not infringe on copyright. In other words, if the training data is available for a person to learn from and if a person could make the same content an AI would and it be allowed, then AI should be allowed to do the same. AI should not (as an example) be allowed to simply reproduce a bit-for-bit copy of its training data (provided it wasn’t something trivial that would not be protected under copyright anyway). The same is true for a person. Now, this leaves some protections in place such as: if a person made content and released it to a private audience which are not permitted to redistribute it, then an AI would only be allowed to train off it if it obtained that content with permission in the first place, just like a person. Obtaining it through a third party would not be allowed as that third party did not have permission to redistribute. This means that an AI should not be allowed to use work unless it at minimum had licence to view the work. I don’t think you should be able to restrict your work from being used as training data beyond disallowing viewing entirely though.

I’m open to arguments against this though. My general concern is copyright already allows for substantial restrictions on how you use a work that seem unfair, such as Microsoft disallowing the use of Windows Home and Pro on headless machines/as servers.

With all this said, I think we need to be ready to support those who lose their jobs from this. Losing your job should never be a game over scenario (loss of housing, medical, housing loans, potentially car loans provided you didn’t buy something like a mansion or luxury car).

We’re going to hold this song back from you and ask for a bunch of your details so you can listen to it once we’ve generated some extra hype. Pretty cool huh?!

The article seems to indicate they are using to reduce the amount of work that have to do in writing prompts, but still have translators review what the AI spits out. I think that’s different to SuperDuo which I believe is mean’t to use AI to be more conversational.

WhatsApp claims to be E2E/not readable by Facebook and to my knowledge, all we have to the contrary is speculation provided you verify the keys on both ends (same as Signal). Facebook might know who you’re messaging but that’s also true for Signal. I’d still 100% trust Signal over WhatsApp given Facebook’s massive conflict of interest, but SMS has been known-bad and collected by the NSA for a decade now. US telecommunications companies also have a terrible reputation for privacy. The only advantage it has over any other platform is portability between providers but even that falls to the side since you can have multiple messaging apps at once.