Meta admits to training LLM AI with terabytes of torrented copyrighted works.

[Quoth]

Originally Posted by DiapDealer

Sorry. But I think what you actually meant to say was that "that also reveals that your interpretation of the legal ramifications of how LLM works does not match my own." Because let's face it. Your interpretation of AI and LLM in general are a bit more philosophical (not to mention semantic) than most's.

No, that's your biased opinion of interpreting what I write.

You are acting as an apologist for an environmental and knowlege damaging scam.

It's cheek to give your twisted interpretations of what I've written. Let people draw their own opinions.

See

Taggart :donor:
@mttaggart@infosec.exchange
The BBC studied the use of its content in LLM output. It found that 91% of the output had mistakes. 51% of the mistakes were "significant."

I will not give up this fight. This technology is detrimental to human knowledge.

https://www.bbc.co.uk/aboutthebbc/document

[DiapDealer]

Originally Posted by Quoth

You are acting as an apologist for an environmental and knowlege damaging scam.

I'm doing no such thing. And it's cheek to give your twisted opinion of what I'm actually saying.

See how that works?

Even you can't deny that that your definitions of these things are often quite literal and semantic. That doesn't make them "correct". It just makes your opinions on them more literal and semantic.

[Quoth]

Thomson Reuters has won a partial summary judgment in a copyright case against shuttered AI firm Ross Intelligence, a decision that disallows fair use* as a defense for training models on proprietary data without permission.
https://www.theregister.com/2025/02/..._ai_copyright/

[* Which is most often the USA law, and there isn't even a fair use defence in many countries]

[SleepyBob]

Originally Posted by Quoth

Thomson Reuters has won a partial summary judgment in a copyright case against shuttered AI firm Ross Intelligence, a decision that disallows fair use* as a defense for training models on proprietary data without permission.
https://www.theregister.com/2025/02/..._ai_copyright/

[* Which is most often the USA law, and there isn't even a fair use defence in many countries]

This court decision specifically does not apply to generative AI.

[DNSB]

Originally Posted by SleepyBob

This court decision specifically does not apply to generative AI.

Without the training, there is no generative AI.

[salamanderjuice]

Originally Posted by DNSB

Without the training, there is no generative AI.

Seems a bit ridiculous not to consider model output in the ruling. You could train a model to output a number between 0 and 1 on how similar a text document is to Harry Potter. Obviously that would need to be trained on Harry Potter but the output would have nothing to do with the content in Harry Potter. Would that require a special license to make?

There are researchers who do similar things. Can't imagine the nightmare of trying to check if you have the proper "training model" license for everything in your corpus. Like here's a paper where they use content from fan fiction sites in machine learning models to glean insights:

https://link.springer.com/article/10...78-024-01224-x

Who is the correct rightsholder here?

[ChrisHistorian76]

Originally Posted by salamanderjuice

Seems a bit ridiculous not to consider model output in the ruling. You could train a model to output a number between 0 and 1 on how similar a text document is to Harry Potter. Obviously that would need to be trained on Harry Potter but the output would have nothing to do with the content in Harry Potter. Would that require a special license to make?

There are researchers who do similar things. Can't imagine the nightmare of trying to check if you have the proper "training model" license for everything in your corpus. Like here's a paper where they use content from fan fiction sites in machine learning models to glean insights:

https://link.springer.com/article/10...78-024-01224-x

Who is the correct rightsholder here?

I think the answer to your question on fanfic, in the US at least, is "it depends."

https://en.wikipedia.org/wiki/Legal_...th_fan_fiction

[jackm8]

Originally Posted by Quoth

Legally as in an obtained licence. But any Publisher is really violating rights of the author to do that without the author's permission. Like if a publisher gets rights for a book, they can't produce an audio book, TV series, play, Cinema, sell prints & models of characters, produce a translation.

But if they do get rights, then they can do all of that, all depending on the contract. There's an interesting case surrounding David Lodges, Changing Places. Author agreed to a screen adaptation, he sold the rights of the novel, then the movie entered production hell, and was never made. The rights were sold for infinite time, so he could do nothing about it.

[Quoth]

An Australian expert view
https://www.theguardian.com/commenti...-akin-to-theft

[Graham44]

Just another reason for met to hate AI. I follow the developments of AI on and off and so far I can't see anything that to me makes AI a good idea, in fact I think it's one of the worse ideas humans have ever had, my hope is it will all implode and just fizzle out (but I can't see that happening).
You will get people say oh but AI can do X,Y,Z - but again it's almost always something humans could do, just AI may be able to do it faster.
It's like a shiny new toy, everyone's fascinated with it and can't stop saying how wonderful it is, but then down the road when the novelty has worn off and the cracks appear, mistakes are made, people will turn away from it and proclaim how it was obvious it was bound not to work

[DiapDealer]

Buggy whips
Tractors

[Quoth]

Not over
https://www.theregister.com/2025/03/..._removal_case/

[Graham44]

Originally Posted by Quoth

Not over
https://www.theregister.com/2025/03/..._removal_case/

Well if Meta get away with this it will be a free for all. It's theft of property, say a person has a question like 'What happened when Europeans discovered America?' and they plugged this into a search engine and an AI search, with the search engine they may come up with a website written by someone that specializes in this subject, now this person may have spent years studying this subject, writing about it, whatever, they then spent the time to write up a paper/ article for the web. By visiting their original content they A) Get credited with the material and B) May get compensation for someone reading the article i.e through AD's on their web page etc - but the main point is they get the credit for their work. If you plug the same question into an AI search, it could well use the same website for its information, spit out the result and you would have no idea that information has come from someone who invested a lot of time into researching and writing up that original article.

The original article would/should also credit any sources it used, but again AI wouldn't credit these, this makes fact checking AI information very hard/ impossible.

Big companies like Meta shouldn't be allowed to steal information, just because people are enamored with the shiny new toy that is AI doesn't make it right.

Although I may be a bit biased as I hate AI with a passion

[John F]

Originally Posted by Graham44

Well if Meta get away with this it will be a free for all. It's theft of property, say a person has a question like 'What happened when Europeans discovered America?' ...

I did the search with Google using your question.

Google summarized and provided links.

[Graham44]

Originally Posted by John F

I did the search with Google using your question.

Google summarized and provided links.

That,s a web search summary not a specific AI generated answer

I tried a free chatGPT and asked it the same question, which it answered and then asked it to elaborate which it did. Zero references in any of it

I then asked it where it retrieved this information, it wouldn't say it just said ''The response provided is based on the context and knowledge programmed into me'' - well that's a lie/ misinformation, it can't all be programmed into it , it must have gathered the information from some external source.