Is it unethical if the AI just learns from the video and then repeats it like a human would?
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple Sued by Three YouTube Channels
- Thread starter MacRumors
- Start date
- Sort by reaction score
Nicky G
macrumors 65816
Nicky G
macrumors 65816
If it violates YouTube’s anti-scraping rules, as well as potentially violates basic aspects of copyright law.
I totally agree with the law suits, sorry but governments literally giving free reign and removing laws for copyright material purely if it is being taken for Ai training, leaves a very bad taste in my mouth and leads a very bad example.. literally invalidating laws to suite giant global megacorps and there profits.
You are talking about global corporations worth multiple TRILLIONS of dollars generating multiple BILLIONS of dollars profit every quarter, literally stealing other peoples work to generate more profits... so damn right they are going to get sued.
Apple tried to shut down Samsung using the same argument and many here defended Apple, which was the equivalent of the You Tube channels, so you're a hypocrite if you supported that action but not these You Tube Channels, although I had issues with Apple trying to claim it invented an oblong shape with round edges, and the colours black and white as they tried to claim in court....
You are talking about global corporations worth multiple TRILLIONS of dollars generating multiple BILLIONS of dollars profit every quarter, literally stealing other peoples work to generate more profits... so damn right they are going to get sued.
Apple tried to shut down Samsung using the same argument and many here defended Apple, which was the equivalent of the You Tube channels, so you're a hypocrite if you supported that action but not these You Tube Channels, although I had issues with Apple trying to claim it invented an oblong shape with round edges, and the colours black and white as they tried to claim in court....
Last edited:
Maybe youtube could give the creators a share? When i use google's ai it will always show the sources that it scraped for information. Those youtube sources could receive some monetary credit even if the searcher does not go to the video
Apple has struck deals with companies to train their AI models in return for Apple paying them millions for access.
I’m curious why Apple is training on YouTube without paying for it. Or maybe they paid Google and these creators should be looking at them instead. Or maybe there’s something in the ToS that allows companies to use YouTube content.
I’m curious why Apple is training on YouTube without paying for it. Or maybe they paid Google and these creators should be looking at them instead. Or maybe there’s something in the ToS that allows companies to use YouTube content.
“Apple's research papers indicate that some of the YouTube videos uploaded by the plaintiffs were used to train its AI models, the complaint alleges.” proof provided by apple, like this?
Have zero merit. It will be thrown out by the judge in a hurry.
If you upload anything to YouTube, it will be used to train AI in some way, especially Gemini. It’s even stated in their TOS. You upload anything, you waive your rights.
Plus Apple hasn’t used any of the AI data yet as Apple’s LLM model is behind schedule due to neural engine still not being powerful enough to run everything on device. So this case has ZERO merit.
Yikes. I just put this into ChatGPT, and it doesn't even agree with your statement. Change your paragraph from an AI company to a bank and a content creator to an old grandma's savings, and see how that feels.
Suggesting that any one creator’s contribution is “small” doesn’t make large-scale scraping acceptable. Scale doesn’t erase individual rights, it just hides them. A bank taking tiny amounts from millions of people without permission is not suddenly lawful. Likewise, taking intellectual property in small chucnks is not automatically justified because the end result is useful. neither is this creating open access to AI anyway. It mainly benefits the handful of companies large enough to scrape, store, and train on the entire internet in the first place.
It does the same things humans do. It looks at data, images, etc, group them into logical chunks, classifies them, and then uses them to create new works. Not much different than how 90% of creators work.
That's not how the law works at all.
You can use any copy righted content if it falls under fair use. Apple profits off the content anything their AI scraps.
Under the current laws worldwide people and companies are allowed to sue anyone who uses their work and it does not fall under fair use. Like if you make a Mickey Mouse plushie and sell it on Etsy Disney can sue you. If Apple uses pictures of Mickey Mouse then Disney can sue Apple.
Courts will eventually need to decide this and US courts will probably get it wrong which will hamstring the US for a few decades while other countries not beholden to Hollywood will zoom forward. Then it will get overturned while we all laugh and realize it was the dumbest thing ever to interpret training a mathematical model as stealing IP.
WarmWinterHat
Suspended
That's nice, l but the courts have already been pretty clear that they're not the same thing. How they work is irrelevant.
One is a human, one isn't. Simple
This is not a correct understanding of AI. There are probably only 0.00001% of concepts where a specific source could be identified. Something super narrow and technical where only 1-2 research papers exist. And in most cases the model wouldn't have learned it because it would require too many parameters to know something so obscure. If the AI learns it there are probably 100s of parallel sources of info.
I think that some of what is driving this is the "fire bad" mentality.
We often want fewer sources from an expert than we do from a novice, depending on the type of claim. So this isn't just about people who consume content and use it professionally without citing sources.
Humans can copy things and we trust them not to or at least to hold them accountable if they do.
Some legitimate problems with this seem to be:
1) if an AI is tricked into re-producing copyrighted materials, it's unclear if it's the person if it is the person who made the prompt or the AI itself, so you have to chase both.
2) AI can potentially be naughty reproduce something at scale without the exposure of the market. tattoo artists violate copyright all the time, but they can only do so much, pre-printed temporary tattoos supposedly are caught in the market place. An AI that reproduces material on demand may never provide an opportunity to get caught but might do it at scale.
We should look at these real problems, rather than just "fire bad"ing it.
We often want fewer sources from an expert than we do from a novice, depending on the type of claim. So this isn't just about people who consume content and use it professionally without citing sources.
Humans can copy things and we trust them not to or at least to hold them accountable if they do.
Some legitimate problems with this seem to be:
1) if an AI is tricked into re-producing copyrighted materials, it's unclear if it's the person if it is the person who made the prompt or the AI itself, so you have to chase both.
2) AI can potentially be naughty reproduce something at scale without the exposure of the market. tattoo artists violate copyright all the time, but they can only do so much, pre-printed temporary tattoos supposedly are caught in the market place. An AI that reproduces material on demand may never provide an opportunity to get caught but might do it at scale.
We should look at these real problems, rather than just "fire bad"ing it.
tazinlwfl
macrumors 6502
If Ethan et al got Adsense for every view that was required for the training data, and the AI companies weren’t allowed to just scrape/download the video and process it as many times offline as they needed, then I’d say Ethan and them got their dues. If their content IS scraped and then trained with off platform, that starts to creep into TOS violation and then the channels definitely were not getting their dues. This is sorta how Ethan got started though, so on these grounds it seems legitimate (even if his overall legitimacy lately has been in question).
"proof provided by apple, like this?" This is a big strong statement. The plaintiffs claim so, but what Apple research papers are the plaintiffs refering to?. We can't know what, if any, from the post.
"the complaint alleges." Again, this means nothing. Trump alleged that Obama wasn't a U.S. citizen. Allegations are not proof of anything.
This post has raised interesting discussions about copyright in general, and in relation to AI in particular, but there isn't enough info about this case in particular in this post to really 'pick sides'. Many comments are, as usual, just influenced by preexisting feelings about Apple.
Black Belt
macrumors 65816
99.9% of YouTubers including the Kleins take their ideas (cough…STEAL…cough) from other people. It’s regurgitating. If anything people like that HARMED AI and made it dumber and inconsistent. Maybe they should be sued for harming AI with their dreck.
yeah, this is the legal language used in court - parties “claim” and “allege”, the court establishes facts.
“this means nothing” does not apply here, since those words don’t have inherit weakness in court setting
chrono1081
macrumors G3
Black_Mage
macrumors 6502a
Unless they know something I don't, Apple doesn't even have A.I. models to train. Apple Intelligence has been powered by ChatGPT and will soon include others, if the rumors are true, that we will get to choose which one we use. This seems like a pointless lawsuit to me.