Training AI models on copyrighted works is "squarely what transformative use is," EFF’s Tori Noble told Reason - these are "general purpose tools" and "exactly the kind of creation that copyright is designed to foster.” reason.com/2026/09/04/doj-says…

reshared this

in reply to 707Kat

@707Kat I guess I never really paid enough attention to notice, I had figured the EFF was on our side and just left them alone to fight my own battles.

It sucks to feel I trusted them... It feels like the ground is crumbling as every major thing I thought was on our side turns out to be terrible (like linux, Debian, and most of FOSS right now).

I really need to start my own org at this point, I don't really know who else to trust.

in reply to Quest 💚

@corigan01 yeah it sucks. I think I only knew because it’s been a subject I have been paying really close attention to.

There are still good people and organizations out there. The EFF does do good regarding surveillance. People are banding together and building moats. But yes, it does make the landscape bleak for current existing organizations.

I think some of the ones that swing gAI will change their mind, as I have seen it elsewhere before. EFF I think not.

in reply to Grutjes

@Grutjes
The EFF has often had a tendency to side with whatever is most profitable for the big tech companies, for example suing the US government to prevent them from seizing consumer data held by Apple, but not raising so much as a whisper about Apple selling that same data to the US government. This may be why some people have referred to the EFF as "the Google lobby" or "the Facebook lobby."

They have, at times, overridden this tendency. I guess this is not one of these times.

in reply to Electronic Frontier Foundation

The irony here is that this is the same EFF that has worked to liberalize copyright law, this lawsuit being no different. Unpopular as AI training may be, I am a bit surprised to see the pushback to the EFF's actions (which honestly looks like a coordinated, corporate-backed effort to shore up support for copyright maximalism), but I get why taking a principled stand is unpopular since it necessarily entails defending unpopular positions occasionally. Even Techdirt defends the practice.
in reply to Electronic Frontier Foundation

I am increasingly feeling like D-Fens in the fast food restaurant asking „ Can anybody tell me what's wrong with this picture?“.
Can anybody explain to me why hip-hop artists and other musicians have been sued into the ground for ten second samples instead of being lauded for creative transformative use?
in reply to Electronic Frontier Foundation

It's so cool to watch every institution I've trusted bend the knee and kiss the ring of pedophile techbro fascists, one right after the other.

Please consider the last donation that I sent to EFF to be the LAST donation that I send.

Effectively Fucking Finished.

This entry was edited (Saturday, September 5, 2026, 1:04 AM)
in reply to Electronic Frontier Foundation

This is NOT all that is going on, EFF account-minder, and I reject your softened and euphemized framing of this issue. This isn't about "training" or "transformation" or any positive purpose, it's about capturing and destroying information, grabbing the books and other works and stuffing them into the machine and then making sure nobody else can use the stuff for their own rackets.

This is a fraud. The entire #tech sector is consecrated to CHEATING. You can't honestly say that there's a positive "transformative" purpose in the GenAI obsession.

There's no reason at all at work. Their thing isn't an "AGI superintelligence" and they know it. When do we stop pretending that this is just kind-hearted software engineers wanting to change the world? They exist, but not anywhere powerful in U.S. business or technology.

#tech
in reply to Electronic Frontier Foundation

Even if creating a generative model is transformative, selling the use of one as a for-profit service to undercut creative artists is preventing the kind of creation that copyright was intended to foster.

But since the generation of statistically similar outputs ensures that the further-derived generated outputs are minimally transformative of the sources, it’s misleading to describe the model itself as transformative.

These services should be required to license their sources.

in reply to Electronic Frontier Foundation

I'm surprised you're bragging about your staff attorney's boneheaded comment.

P.S. What are you patting yourself on the back for? Having a your staff attorney quoted in an article? I used to think EFF was cool but shameless self promotion is the opposite of cool. It's lame and gross.

This entry was edited (Saturday, September 5, 2026, 4:27 AM)
in reply to Electronic Frontier Foundation

The media in this post is not displayed to visitors. To view it, please go to the original post.

Maybe these stupid #AI #Clankers can stop this stupid #DDoS attack on my server then?

If you have taken this hypocritical take on copyright, then you need to get your head out of you're ass, because if I did the same thing these companies are doing, I'd be sued into hell for it.

Translation: Someone at the #EFF is definitely getting fired over this stupid post.

This entry was edited (Saturday, September 5, 2026, 6:04 AM)
in reply to Electronic Frontier Foundation

What kind of ass backwards take is this? Does training models on copyrighted materials bring any additional money to the creator of the copyrighted work used in said training? No. If there's no monetary benefit to the original creator of a work, it should not be used in training.

I'm so disappointed in you all for backing this horrible take.

in reply to Electronic Frontier Foundation

a horrible opinion that screws creators. Unbelievable that you would not only support mass copyright infringement, but seemingly with enthusiasm. You've officially helped tip the scales in favor of billionaire robber barons and likely alienated a large percentage of your supporters.

Are you counting on large donations from the supremely entitled, inhuman, hoarding scumbags?

This entry was edited (Saturday, September 5, 2026, 7:18 AM)
in reply to Electronic Frontier Foundation

So you think it's somehow "fair use" despite it failing pretty much every test once the marketing newspeak is stripped away?

  • They are barely transformative. They are at best a lossy statistical compilation of multiple works, and any additional value is generally from those other works.
  • These machines are 'trained' with entire works, and their whole purpose is to "learn" (read: attempt to copy) the substance, so fail the "amount and substantiality" test quite solidly.
  • These machines are actively depriving the original authors of income by pushing them out of the market after copying their work, and this is their stated intent.

That last one is particularly damning: They have openly stated that they want their 'product' to push the very creatives they copied from to "train" it out of the market, which is one of the reasons we have copyright: If everyone's work is immediately hoovered up and shat out worse by a slop-machine, they have no incentive (or ability) to create.

This entry was edited (Saturday, September 5, 2026, 1:26 AM)

reshared this

in reply to Electronic Frontier Foundation

Saw this coming. EFF has been on Big Tech's side for every issue except surveillance capitalism and censorship.

Well, this sucks. Thanks for siding with fascism, EFF. I'd suggest abandoning Open Source Software as a cause of yours as well as the Creative Commons, because if AI is going to have your backing, you can't support either of those. (You have lawyers, you should know why.)

(Edit: this was me chiming in, nothing more. I'm muting this thread.)

This entry was edited (Saturday, September 5, 2026, 6:49 PM)
in reply to Tock

Sensitive content

@Tock
This entry was edited (Saturday, September 5, 2026, 8:22 AM)
in reply to Electronic Frontier Foundation

so when these things mulch gpl code and shit it back out in another project without attribution that’s alright then? This is a profoundly shitty take that devalues the work of everyone so a handful of corporations can loot other peoples work without compensation, EFF of all organizations should be better than that.
in reply to Electronic Frontier Foundation

I remain unconvinced.

As a layperson, I’ve been told that the point of copyright has been to make sure the writer and their publisher can earn a livelihood, so they may live to write another days

I know if I were to consistently copy or distribute copyrighted articles in such a way that it affects these companies bottom line to the point where they downsize, they’d come for me with an arsenal of lawyers.

So there are consequences for us small people.

But along comes OpenAI and the gang, and because they transform these copyrighted works into a model, it’s suddenly alright if they don’t license it even though they actually replace a whole numbers of writers?

That doesn’t add up to me.

I don’t buy the hand waving in this article about “it’s not replacing jobs” either. It shouldn’t, but it does, right now, because management isn’t always smart.

I think copyright is flawed and a bandaid for a not-working economic system, but you can’t rip off the bandaid before you fixed the system.

So that’s a no from me.

Em reshared this.

in reply to Electronic Frontier Foundation

speaking as a card-carrying paid up #EFF member… fuck off with this shitty take 😡🤬

if you’re gung ho on copyright, how about something more defensible like “we believe that (US) copyright laws are a net positive but in the light of wholesale unpaid slurping of human creativity and labour for billionaire profits, clearly they’re broken and need to be fixed – here’s how we plan to fight this”

JFC, do better 🤮

#EFF
in reply to Electronic Frontier Foundation

So, uh, you taking donations from the Slop companies now? Is this why you're dropping toxic takes like this?

They're literally destroying the commons by scraping web servers to death, they're burning the planet with this stampede to build data centers and polluting water, and then their poorly designed models aren't even any good at what they claim to be able to do. They introduce bugs, give bad answers, and record incorrect information in medical transcripts!

But sure, fight for the big guy as he puts his boot on the necks of the rest of us.

This entry was edited (Saturday, September 5, 2026, 4:41 AM)
in reply to Electronic Frontier Foundation

fair use is not for profit. What they did is. Libraries are fair use ( yes, including shadow libraries and torrents) Fair use is a human tool, for sharing knowledge and art without commercial consideration. Fair use has nothing what-so-fucking-ever to do with what the corporations do.
in reply to Electronic Frontier Foundation

extremely obvious outcome is extremely obvious


Training AI models on copyrighted works is "squarely what transformative use is," EFF’s Tori Noble told Reason - these are "general purpose tools" and "exactly the kind of creation that copyright is designed to foster.” reason.com/2026/09/04/doj-says…

in reply to Electronic Frontier Foundation

you’re supposed to be an organisation that stands up for our digital rights. Our rights include not having work taken wholesale and regurgitated by for profit companies without proper licensing and proper attribution - in most cases actually even costing us money as these bots steal bandwidth meant for humans.
in reply to Electronic Frontier Foundation

The media in this post is not displayed to visitors. To view it, please go to the original post.

I have never been a *recurring* donator, I renewed annually whenever I had some money left. I do this since 2011.

VERY much rethinking this for this end of the year.

Your work is sorely needed, but AI certainly is not one of the things I think you should put my donations towards.

in reply to Electronic Frontier Foundation

You know your take is good and you are on the right side of history when you are siding with corporate interests in a puff piece from the Koch funded Reason outlet, and in great company, with their long history of defending and platforming apartheid defenders, big tobacco, climate change deniers, big oil, holocaust deniers, covid deniers...
in reply to Electronic Frontier Foundation

cmon EFF, what corporate contributions are you defending with that conclusion? It is resoundingly false. Your position should be the individual creator not some corporate monolith scraping the globe for all known human knowledge with the goal of regurgitation without citation. It is the largest copytheft industry in history. The courts are wrong on this. You're wrong on this.
This entry was edited (Saturday, September 5, 2026, 10:43 AM)
in reply to Electronic Frontier Foundation

I am in no way a supporter of how copyright laws are used at times especially to attack remixes but the LLM-case seems extremely clear: Fair use clearly limits commercial usage, especially one where you directly compete with the original work.

WTF are you going on about saying running a business like OpenAI under fair use is allowed?!? If they gave away the model for free then sure, this could be argued, but they are literally making money off of stolen works.

This entry was edited (Saturday, September 5, 2026, 12:52 PM)
in reply to stux⚡️

@stux
I have been a longtime supporter of the EFF, and hope to remain so

That said, after this statement of theirs, I have no reason to continue to do so

I fully understand the need to make sure that the right of "transformative use" is maintained.

These LLM's are not transformative. Time and again it has been shown that copyrighted works are able to be dispensed *verbatim and in full*

These companies do not credit source, and (try to) claim copyright on these materials themselves

in reply to Electronic Frontier Foundation

so how the f does that square with privacy and free speech????

Stealing is a violation of privacy, stealing our words is a violation of free speech. No one is being compensated. Rare books are being destroyed. In some cases original authors are being denied the rights to their own work because an AI stole it from them.

You can not say this is not stealing.

Absolutely you have violated your own purpose in existing.

#EFF

#EFF
in reply to Electronic Frontier Foundation

and what about verbatim-enough output that would certainly be considered copyright violation if a human hand-typed it? Even if (and I'm far from sure i agree) training a model in and of itself is legal, it's output could still violate copyright. So either this doesn't say much, or copyright is effectively dead.

Additionally, the article contains claims about prohibitive costs if the trainers need to get permission from all sources. Those costs are currently externalized to those sources, so maybe that would he a good thing? The scrapers are literally killing the web right now.

in reply to Electronic Frontier Foundation

Thats a scumfuck move and if you where doing your job you'd be pushing this fight so that any ai code that came from a model trained on GPL/agpl code must be agpl'd.

Fuck you guys, you lost the plot. Please fall down a shit-covered spike pit.

This entry was edited (Saturday, September 5, 2026, 10:34 AM)
in reply to Electronic Frontier Foundation

They aren't "general purpose tools," this *absolutely* isn't anything remotely like the creation that copyright is designed to foster, and it definitely isn't a transformative work. I get that she's a lawyer by trade and training, and therefore lacks the background to understand what "AI" is, so I have to wonder who she's been working with to have developed such an enormously skewed idea of what this technology actually is and how it's created.
This entry was edited (Saturday, September 5, 2026, 12:26 PM)
in reply to Electronic Frontier Foundation

Yes and reading a book is transformative to my knowledge and mind, so just give me all the copyrighted works. Does anyone dispute that a human reading those works makes a human better/smarter/more creative?

I'm not a huge fan of copyright holder lawsuits. However, AI companies are stacking cash, while standing on the heads of all of human history. It's not a public works project for the betterment of society, it's not a library you can freely use. It's not even like the beginning of the internet, where infrastructure and networking allowed collaboration and content creation.

Love the EFF but I don't think I'd help the AI overlords in this. Spend the time on something better than this.

Heaven forbid that the courts set a precedent that OpenAI pay for content they use for training purposes. Maybe they could just let them use torrent like Meta did already. I mean it's accepted use right. (/s)

in reply to Electronic Frontier Foundation

I'm so confused. In front of me is Cindy Cohn's "Privacy's Defender". The opening pages read:
"(c) 2026 Massachusetts Institute of Technology
All rights reserved. No part of this book may be used to train artificial intelligence systems or reproduced in any form [...]"

What kind of a pro-slop-technofascism stance did you assume over night? This must be a joke, right? Clear up this mess please.

This entry was edited (Monday, September 7, 2026, 5:02 AM)
in reply to Electronic Frontier Foundation

A technology only available to ultra-wealthy corporations that is built through copyright infringement is A-OK, but normal people can't have a a 4 second clip of Disney in the background of a video without getting sued.

Is that really what you're saying? I just want to be clear before I cancel my 25-year membership.

in reply to Electronic Frontier Foundation

Wow, the irony is palpable in this one. This is the most "technically correct but in unintended way" take we've seen in a month.

Yes, this is *precisely* the kind of "creation" copyright is *designed to foster*. Corporation-centered, worker-oppressing framework to extract profit from others' labor and handwave away any complaints through legalese bullshit and arguing for fake moral superiority, when the reality is that multi-billion companies use copyright as a tool for oppression. genAI is just another iteration of "laws for thee but not for me", alongside many other oppressive things.

Shame on you.

in reply to Electronic Frontier Foundation

Hey, uh...
So if I set up a pantograph and use it to copy the contents of an entire book by tracing over the pages, and the pantograph is poorly-designed so it results in lots of errors, but the book is slightly larger and therefore distinguishable from the original by size alone, is that a transformative work? Especially if I sell it as an alternative to the original?

I'm just wondering where the limits apply.

in reply to Electronic Frontier Foundation

I'd like to kindly remind everyone in this thread that copyright is the legal system that has allowed everyone from Apple to Facebook to Twitter to Nintendo to Android to all build walled gardens around their platforms and slowly enshittify everything until the whole internet is as terrible as it is today.

Copyright law has been the method through which all of that has occurred. Copyright is a mechanism that enables powerful corporations ("rights-holders") to acquiesce even more power.

The fact that we're in a temporary moment where a few massive companies have found a winning strategy in subverting copyright, this does not mean that copyright isn't the mechanism destroying the entire internet.

youtube.com/watch?v=4EmstuO0Em…

in reply to Electronic Frontier Foundation

Swilling the tech bro kool-aid. So do you also agree with Altman that intelligence should be a utility that we all pay him to access?

Or do you have a 30,000-foot theoretical stance to deflect attention from what's really going on here?

in reply to Electronic Frontier Foundation

AIANAL, but I find obvious that using copyrighted content to create content that competes with the copyright holder is far from being 'fair use'.

The main reason I am not contributing to open source is I don't want my code to be used to be used by AI. I don't mind if code derived from my contributions were to be published under the same license, but that's clearly not the case with AI.

Stop helping big corporations destroy the commons, please.

in reply to Electronic Frontier Foundation

> Training AI models on copyrighted works is "squarely what transformative use is," EFF’s Tori Noble told Reason

People are very upset about this. But it's not a simple 'pro-AI' vs. 'anti-AI' question. It's a complex legal one about where the boundary lines between fair use and copyright violation ought to lie. One can be vehemently opposed to #MOLE Trainers (as I am), without necessarily wanting the copyright monopoly strengthened in the way Big Content is pushing for here.

(1/?)

#AI

#AI #mole
This entry was edited (Sunday, September 6, 2026, 2:14 AM)
in reply to Strypey

It's analogous to the link tax issue. Would I feel sorry for DataFarmers if they had to pay individual journalists a royalty whenever their work was linked on DataFarming platforms like FarceBook or Xitter? Not for a second.

So do I support link taxes as policy? Also, not for a second. Not because of their affect on the DataFarmers targeted, but because of the precedent created. If linking required a copyright license, and perhaps a royalty, that would fundamentally break the open web.

(2/?)

This entry was edited (Sunday, September 6, 2026, 2:14 AM)
in reply to Electronic Frontier Foundation

I thought this was a quote from the DOJ and that EFF was (rightfully) calling them out. EFF needs to get on the right side of history or forget about any future donations from me. The only thing "transformative" about laundering copyrighted materials through a plagiarism machine is how it turns other people's work into money for billionaires. This is an absolute disgrace and an unforced error.

Edit: you know what, I cancelled, I'm not going to act like my threat means anything to you. What a waste. When the institutions that are supposed to be standing up for freedom and fairness start talking like and supporting the opposition you can't make idle threats. I hope someone else stands up where you fell down.

This entry was edited (Sunday, September 6, 2026, 2:00 AM)
in reply to Electronic Frontier Foundation

lawyers gonna tell you how they see it. Doesn't make it right or wrong.
I do not think fair use applies here. This also doesn't touch native rights at all. One corner of law in very narrow interpretation might be correct, it wider context it is still wrong.
This is usually where new laws get written tho.
in reply to Electronic Frontier Foundation

Ok, so I’ll download copyright material to train my own actual intelligence? That’s fine right and fair use? Nobody will sue me for breach of copyright, right? RIGHT?

My goddess give your head a wobble. Supporting mass theft of people’s work to feed to the Great Fascist Hallucination Machine? That’s probably the most stupid thing I’ve heard. To hear it from the EFF? Beyond stupid.

in reply to Electronic Frontier Foundation

Oof. Not the take I would have expected from you.

The fact that a well crafted prompt can pull near identical data from the training source shows this is not the case.

Touch grass, @eff, I think you have spent too much time online.

This entry was edited (Monday, September 7, 2026, 9:44 PM)
in reply to Jim Jones

@GreatBigTable

This is simply not true in the general case*. The proof is simple and based in computer science and maths:

A great LLM to host locally is Qwen 3.8 27B. Quantized to 4 bit weights the size of the model is ~14GB on disk.

All of human knowledge cannot be compressed down to 14GB and then decompressed again (see Shannon's theorem). Thus LLMs do not work by storing training data.

*) When training has _failed_ and cause what's known as "overfitting" too much training data is stored close to verbatim in the model. This is unwanted (wastes model space) and are extreme outliers model creators work actively and successfully to make sure doesn't happen.

@eff

This entry was edited (Sunday, September 6, 2026, 12:09 PM)
in reply to Electronic Frontier Foundation

It's not 'transformative' when these techbros just decide to throw copyright of the window now when it suits them, while for decades they used the same copyright to jail users and do all sorts of harm. This is not 'transformative' it's called being deliberately oppressive and using Law and Technology to propagate oppression. I am also disappointed in EFF. Looks like we need to rebuild our organizations, none of them seem to protect us, the humanity, the earth and our public commons.
in reply to Divya Ranjan Pattanaik

@divyaranjan There's a world of difference between "training is fair use" and abandoning copyright protections entirely.

The copyright office has different guidelines for training an LLM and distributing the output of an LLM. Rights holders of training data might need to licence outputs before they can be distributed without infringing.

I'm sure the EFF is taking this stance so that individuals can train and use LLMs for good ends both now and after the bubble pops. Sometimes laws and legal principles protect both large organisations and individuals, both to good ends and ill.

Even if "training is fair use" we can still fight the generative AI abuses, especially of the companies commercializing it while trying to shirk responsibility, both through legal and labor efforts.

This website uses cookies. If you continue browsing this website, you agree to the usage of cookies.