⚡️🌙: — Subium • Bluesky web client

In 2020, under lockdown, I was in calls with some amazing lawyers to push back against the RIAA's ridiculous claim that youtube-dl was illegal. github.blog/2020-11-16-s... That is to say, if you are cheering them on now because you hate AI, you forget what copyright maximalism is. Just block me.

Standing up for developers: youtube-dl is backgithub.blog Today we reinstated youtube-dl, a popular project on GitHub, after we received additional information about the project that enabled us to reverse a Digital Millennium Copyright Act (DMCA) takedown.

I'm seeing young people on this website and on Twitter posting pro-copyright "you wouldn't steal a car" bullshit from the 2000s because they're mad that a computer that looked at a lot of arrangements of pixels can render new arrangements of pixels. Dark times are ahead.

Echoing this person.

wow what if copyright helps big corporations more actually!

there are no examples of this

how is 'Only """prompters""" can compete in the art market' any better? why do you suppose there's no market or appetite for art actually made by humans? if linear algebra can do fair use then why can't I horizontally flip an episode of seinfeld, upload it to youtube, and expect to not get DMCA'd?

My problem with the big pro copyright companies and the big generative ML companies are almost exactly the same - they both try to take all profits for themselves, away from people who create their own works, making hollow arguments about "the best for the industry"

I don't want to limit fair use, and I don't believe it's needed either, you can deal with exploitative gen ML in other ways. Gen ML should be legally equivalent in terms of liability as if you gave the same prompt to a human artist with a similar result. Nothing about the fancy math changes this.

You don't have to expand copyright by declaring that if a human artist would be sued for infringement and lose them the gen ML system is exactly equally liable for creating an equivalent mimic of existing work. Learning is protected, original creations / transformative works remain safe, etc...

But of course then ML companies have to check outputs for similarities to their samples and track their copyright status and they REALLY don't want to do that. It's just a few companies like Adobe that even bothers ensuring their samples are licensed such that replication isn't a legal problem

I agree that we already have all the legal tools in place! And we should use them! But why would that be the responsibility of the companies and not the end user? Adobe doesn't currently have to make sure you're not drawing Mickey mouse with Photoshop. That's my responsibility as the end user.

Regular Photoshop isn't built from datasets of copyrighted media, output is driven fully by end user input. Unless the user is knowingly entering prompts about copyrighted stuff they might not be able to tell when the ML tool replicates stuff from samples - but the tools can build in detection

AI isn't "fair use" it's fancy plagiarism.

The corporations ALREADY control the commons. Fanartists are already treated like garbage and they MAKE their art. All your doing is defending the rights of theives to muscle out independent artists by using technology built on THEIR WORK without their knowledge or permission.

Like who do you think you're ACTUALLY helping by excelerating the plagarism arms race???

Post