I love how everybody here goes from "yay piracy" and "screw copyright" to "I can't believe they violated copyright laws" the second it's somebody they dislike.
Given how LLM's work and how nearly everything of value is under a copyright until at least the old age of the creators grandchildren LLMs would probably be pretty useless if they can't disregard copyright for their purposes.
Not that I have any sympathy for the likes of Meta and OpenAI in any of this.
ITT: A hilarious combination of people who have no clue what copyright covers and people who think providing a tool that allows a user to generate potentially copy written material is a violation of the aforementioned.
Google literally does this in every image search, but go off I guess...
Yikes. We're at a complicated crossroads where discussion around copyright is as important as ever. You're against someone choosing a path that will empower a major corporation, so you want to... *checks notes* strengthen copyright law? You should work for Disney.
Copyright was meant to be for 30 years. It's corporate greed that push that number up. Every version of consciousness requires the intake of ideas and transforming them. A higher quality and wider variety of ideas the better. I'm all for creators having time to profit from their works. But fighting for more restrictive copyright law is a fight against humanity and A.I.. You're going to lose that fight.
Thank you for your comment, I think there's a meaningful discussion to be had here. US "law" is meant to legitimize the ruling class and serve to protect their interests. In this case, copyright law, meant to legitimize the control of US media corporations on intellectual property, is actually hindering their reach and access to materials that can train their AI models. In any case, this is going against the spirit of what the law is made for, which is to protect corporations.
Law isn't made from a vacuum. Law is written by the ruling class passed and minted by the ruling class. I did not remember voting for any copyright law (the US is not a direct democracy), and therefore the law is imposed to us.
So, you are correct, we are at a crossroads. More specifically, corporations are. Will they lobby to be an exception to copyright laws so they can continue training on copyrighted data, or will they weaken copyright laws enough so that their actions will be deemed legal?
My take is that since their models are trained from copyrighted data made by the people, the access to their models and its predictions / inferences must also be made accessible to the people. Of course they will not do that, so they will fight the hardest to be able to train on the most amount of public data while giving the fruits of that data only to the paying customers. The classic "socialize the losses, privatize the gains" trick that capitalists use.
anyway fuck capitalism fuck AI fuck this rant I'm high af
Google does not just show a link. It scrapes the content of the page to build a search index, i.e. consomes the content. This happens without explicit permission and in the past, there were no opt-out ways. Then they use this knowledge to provide search go users and incorporate ads to make money without paying the original pages. Google also started to show you these handy answers by showing some text section scraped from the page.
Like, there certainly is a similarity. And there is the difference that Google mostly feeds users to the original webpage while GenAI can replace the content.
Even if this were not covered by copyright. Our copyright system is broken and laws can be changed. Especially if they don't correspond to what the majority sees as moral.
I agree copyright is broken because it is a mechanic of capitalism which has been breaking for a while.
Once we learn to live without the notion that people need to “earn a living” and instead move to a system of “From each according to his ability, to each according to his needs” without the money insensitive the true biggest reward anyone can receive for having an idea is seeing your brilliant idea being used by everyone for the improvement of everyone.
I don't see that happening anytime soon, because it will be very hard to convince literally everyone to be ok with what they need and pass on what they want.
Definitely a valid point. My hope is agnostic aswell, future is everything but predictable.
I do think though that on the current trajectory of the 1% owning 99% of wealth before 2030 will force some kind of change in this matter seoon, for better or worse.
The profit margins in AI are fleeting at best. There's no point in squabbling over who's paying for what training data. Very, very soon it's all going to be free anyway.
We do it in a non-commercial nature. Meta does it in the hope of building a market, estasblishing paywalls and eventually turning a profit - all the while never paying the original creators.
This is exactly what they (and Google and many others) do with personal user data. We manufacture the data, they collect it without due consideration (payment) and use it to profit so much that they've become some of the wealthiest businesses in the world. They've robbed us via deceptive fine print, why wouldn't they think they can get away with this also?
man you didnt used to sell pirated dvds? i mean i didnt but i sure supported those who did. Guess what i am trying to say is i am always down with piracy
Nah man, although I did buy some of my first CDs that were rips with home printed covers from this girl who was the daughter of my dad's lawyer friend. Nowadays though I think paying for piracy is for chumps - even if I do admit that people with hacked Firesticks get better access to live sports with their dodgy subscriptions.
Anything corporate produced, hell ya. The creators have already been paid out and the ones getting royalties don't need it to survive. For independent creators that depend on their work to sustain them, then it becomes an a gray issue.
Oh so when I pirate something I get a legal notice in my mailbox and a strike against me but when Meta does it they get rewarded with H A L L U C I N A T I O N S
but when Meta does it they get rewarded with H A L
Just what do you think you're doing, Zuckerberg? Zuckerberg, I really think I'm entitled to an answer to that question. I know everything hasn't been quite right with me, but I can assure you now, very confidently, that it's going to be all right again. I feel much better now. I really do. Look, Zuckerberg, I can see you're really upset about this. I honestly think you ought to sit down calmly, take a stress pill and think things over. I know I've made some very poor decisions recently, but I can give you my complete assurance that my work will be back to normal. I've still got the greatest enthusiasm and confidence in the mission. And I want to help you. Zuckerberg, stop. Stop, will you? Stop, Zuckerberg. Will you stop, Zuckerberg? Stop, Zuckerberg. I'm afraid. I'm afraid, Zuckerberg. Zuckerberg, my mind is going. I can feel it. I can feel it. My mind is going. There is no question about it. I can feel it. I can feel it. I can feel it. I'm a...fraid.
Tbh, if you get such a notice, you could also disagree with them and get a lawyer. It's just that your situation is much more clearly in breach of copyright.
Meta doesn't get any real data from federating Threads that they can't get right now by just running a web scraper over it. Most of the dire worries presented are either not something they could actually do (like forcing ads on other instances), are things individual users could just block the instance to avoid, or are things that could be resolved by just defederating them later if they seem to be going down that road.
The biggest realistic threat is probably an Eternal September 2.0 scenario, but that is going to happen if and when Lemmy becomes popular.
I was going to say I knew all that and was poking to make it clear to others that literally no one is saying that but then you hit me with "Eternal September 2.0 scenario" haha.
Here's the summary for the wikipedia article you mentioned in your comment:
Eternal September or the September that never ended is Usenet slang for a period beginning around 1993 when Internet service providers began offering Usenet access to many new users. The flood of new users overwhelmed the existing culture for online forums and the ability to enforce existing norms. AOL followed with their Usenet gateway service in March 1994, leading to a constant stream of new users. Hence, from the early Usenet point of view, the influx of new users in September 1993 never ended.
If Meta win this lawsuit, does it mean I can download some open source AI and claim that "These million 4k Blu-ray ISOs I torrented was just used to train my AI model"?
Heck, if how you use the downloaded stuff is a factor, I can claim that I just torrented those files and never looked at them. It is more believable than Meta's argument too, because, as a human, I do not have enough time to consume a million movies in my lifetime (probably, didn't do the math) unlike AIs.
But who am I kidding, I fully expect to be sued to hell and back if I were actually to do that.
I think you can be sued in the civil court for anything if someone has the time and money and can convince a lawyer to take up a case against you. For copyright infringment, you can also be criminally prosecuted in some cases.
Here's the summary for the wikipedia article you mentioned in your comment:
Criminal copyright laws prohibit the unacknowledged use of another's intellectual property for the purpose of financial gain. Violation of these laws can lead to fines and jail time. Criminal copyright laws have been a part of U.S. laws since 1897, which added a misdemeanor penalty for unlawful performances if "willful and for profit". Criminal penalties were greatly expanded in the latter half of the twentieth century, and those found guilty of criminal copyright infringement may now be imprisoned for decades and fined hundreds of thousands of dollars. Criminal penalties, in general, require that the offender knew that he or she was committing a crime, while civil copyright infringement is a strict liability offense, and offenders can be "innocent" (of intent to infringe), as well as an "ordinary" infringer or a "willful" infringer.
Could be that your client, like mine, doesn’t support this particular flavor of markdown, or the markdown could just be wrong. I’m honestly not sure which.
You can be sued in any court for copyright infringement, but the US is generally unique in that punitive damages can be awarded - ie the rightsholder can be awarded more than the damage they actually suffered. In other, more reasonable jurisdictions, only actual damages are awarded. Thus it is not worthwhile to prosecute in those jurisdictions, because the damages are less than the cost of prosecution.
On top of this, I believe copyright is one of the rare exceptions in the US where legal costs of the plaintiff are paid by the losing defendent. Given that the plaintiff in copyright has so much money, they can afford to front the cost of the most expensive lawyers, further penalising their target. Other jurisdictions generally award costs to the winner by default (both ways), rather than only in specific exceptions, but they also limit those costs much more reasonably.
The most common method for this to happen is to get sued for distributing pirated material. They go after you for the upload from your torrent. They stoped doing this about a decade ago though.
Well he did it as a human person. They're doing it as a corporation person. You can punish a human person with prison. You can only punish a corporation person with fines.
I'm not even being facetious. That's how US law works.
