> But if they take it, then package it, then make money? That is different

But still, also legal.

You can't copyright a recipe itself, just the fluff around it. It is totally legal for somone to visit a bunch of recipe blogs, copy the recipes, rewrite the descriptions and detailed instructions and then publish that in a book.

The is essentially the same as what LLMs do. So prohibiting this would be a dramatic expansion of the power of copyright.

Personally, I don't use LLMs. I hope there will always be people like me that want to see the original source and verify any knowledge.

I'm actually hopeful that LLM reduction in search traffic will impact the profitability of SEO clickbait referral link garbage sites that now dominate results on many searches. We'll be left with enthusiasts producing content for the joy of nerding out again. Those sites will still have a following of actually interested people and the rest can consume the soulless summaries from the eventually ad infested LLMs.

▲

bbarnett 2 days ago | parent [-]

It may be legal in your jurisdiction, but I think this is a more generic conversation that the specific work class being copied. And further, my point is also that other parts of copyright law, at least where I live, view "for profit copying" and "some dude wanting to print out a webpage" entirely different.

I feel it makes sense.

Amusingly, I feel that an ironic twist would be a judgement that all currently trained LLMs, would be unusable for commercial use.

▲

shkkmo 2 days ago | parent [-]

> other parts of copyright law, at least where I live, view "for profit copying" and "some dude wanting to print out a webpage" entirely different.

I don't know what your jurisdiction is however through treaties, much of how USA copyright law works has been exported to many other countries so it is a reasonable place to base discussion.

In the USA commercial vs. non-commercial is not sufficent to determine if copying violates copyright law. It is one of several factors that is used to determine "fair use" and while it definitely helps, non-commerical use can easily infringe (torrents) and commercial use can be fine (telephone book white pages).

> a judgement that all currently trained LLMs, would be unusable for commercial use

I sure hope not. I don't like or use LLMs but I also don't like copyright law and I hate to see it receive such an expansion of power.

▲

bbarnett 2 days ago | parent [-]

> much of how USA copyright law works has been exported to many other countries

I'm not blaming you for bringing it up, however I did make it clear that I was speaking of a different jurisdiction. And yes, of course you're right, it's always a "big deal" when trade negotiations come up.

Canada has multiple different things in play to protect the individual. The non-profiting dude. Fair use is one, far expanded. Notice-and-notice is another, which currently means you have to pay to send an 'infringed' notice to people, as a copyright owner. Damages are also capped, at an amount that makes legal action untenable for most. And the bar of proof is significantly higher.

And that's for torrents.

For years we've had things like "you pay a tiny tax on hard drives", but then "that means you've already paid for anything you'll ever copy" and the tax goes into a fund to pay Canadian artists. While this may seem strange, it's one solution we've had to help keep art alive, but also not punish the average citizen with crazy law suits, and insane attacks from massive law firms.

Essentially, we don't let the US bully us into agreements which are massively harmful to our citizens.

But back to the LLM side. I see the current situation a weakening of copyright law, a massive one. And not for the average joe, but instead for the most commercial of entities.

I want copyright law, in some circumstances, to be weakened for people. Not companies. They get to pay artists. Creators. Developers.

And of course, there'd be no GPL without copyright law. So while I agree for individuals, especially in the US, copyright law is very annoying and a problem? Let's again focus on what I'm saying.

It currently isn't and doesn't have to be an absolutely

You can and we already have, as we've both discussed, different outcomes for copyright. EG both for fair use and breach outcomes, for corporations/for-profit and just some person. So let's stop talking about copyright stronger/weaker as a generic, and a specific.

I support weaker outcomes of breach, and enhanced fair use for people.

I support stronger outcomes of breach, and so forth for companies.

Further, I support sliding scales too. A one person youtuber isn't the same as a 10B company. A person playing parts of one song in their video for a few seconds, as a one person corp, isn't the same as an entity scanning all of humankind's knowledge and laughing in our faces.

Huge differences of scale and scope.

Look at it this way. Some of these companies have downloaded torrents. If a person did what they did, they'd receive billions in fines!!

Yet they're getting a lesser outcome, as in freaking nothing.

It's the wrong place for copyright weakening.

▲

shkkmo 2 days ago | parent [-]

> I see the current situation a weakening of copyright law, a massive one. And not for the average joe, but instead for the most commercial of entities.

You gonna have to explain this in more detail because it isn't clear to me how you justify this claim. What exactly is being weakened? In what way?

> Some of these companies have downloaded torrents. If a person did what they did, they'd receive billions in fines!!

The one I am assuming you are referring to is Meta, and they are getting sued. They arguably should also be facing criminal charges too under current law.

> Yet they're getting a lesser outcome, as in freaking nothing.

That court case hasn't finished and that doesn't have anything directly to do with LLMs but with our legal system and power/wealth imbalances.

> And of course, there'd be no GPL without copyright law.

I personally strongly prefer MIT to GPL. GPL sort of makes sense as a reaction to copyright law but I don't think GPL justifies the existence or state of copyright law.

> Further, I support sliding scales too.

What does that mean? Just the fines / judgements because along with having to pay, the activity itself must be stopped.

If copyright only prohibited larger entities from copying, it would be less onerous and would make copyright more tolerable, but I don't think that would solve the AI training issue in any way and seems like a tangent.

> an entity scanning all of humankind's knowledge and laughing in our faces.

Knowledge is not copyrightable. If you want to stop this, expanding the power of copyright to make learning/knowing something an infinging activity is one of the worst possible ways to go about it.

	▲	rangerelf a day ago \| parent [-]
		> The one I am assuming you are referring to is Meta, and they are getting sued. They arguably should also be facing criminal charges too under current law. I think your assumption is falling too short, it's not just Meta, it's OpenAI, it's Anthropic, it's Google, and Microsoft, and others. Like you said, the court case hasn't finished, but there's meddling from the Whitehouse already; I really doubt there's going to be any fair play in this case.