Discussion
Under what conditions should copyright-protected content be used for AI training?
🗣️We know this is a topic many of you feel strongly about.
We want to hear your views on issues related to the use of copyright-protected content for training AI models, for research, and more.
📝Consultation open until 3 November 2026.
Hi @jjj, rest assured, we read your comments and do our best to reply🙂
@EUCommission @jjj ...but do mastodon comments actually get carried through to decision makers the same way responses on the website do?
Hi @LionsPhil, comments on social media are useful in furthering the debate amongst people online, and the engagement generates more visibility for the post. We do read our comments and report back to policy makers on a regular basis, but taking part in the consultation is the best way citizens can affect EU policy 🙂
@EUCommission Oh, hi! Good to know, happily surprised to hear that! Thanks for correcting me!
@jjj @EUCommission That said, I think it's safe to assume the the social media team of the Commission is separate from the decision-makers conducting this survey, so you're still correct that rights holders should fill out the (rather technical) questionnaire rather than just voice their grievances on Mastodon.
@brekke @jjj Indeed, it is important to actually take part in the open public consultation, as this is one of the best ways in which citizens can affect EU policy.
Comments on social media are useful in furthering the debate amongst people online, and the engagement generates more visibility for the post. We do read our comments and report back to policy makers on a regular basis, but as you said, taking part in the consultation is the best way to ensure your voice is heard.
@jjj @EUCommission they read, and even answer. Which I find quite awesome. That's so rare. Amazing social media team.
Hi @gunstick, we appreciate it, thank you very much😍
@EUCommission When they begin to pay fair royalties for the usage of the content they don’t own.
Digitally signed license agreements and auditable fair usage of the non-owned content and legal system of tracking of the royalty payments.
Make them pay, too.
@EUCommission
When the compensation is fair enough to avoid me to go bankrupt
@EUCommission
Copyright is Copyright. If the AI Company wants to buy the stuff, a custom agreement with a custom price could be made between the copyright holder and the ai company if the copyright holder wants to do this.
@EUCommission A hefty monthly royalty fee.
They might steal, I mean download, once, but use it daily. So, that means paying.
And as it will be impractical to specify the use, the monthly licence fee is an appropriate solution.
The core problem is copyright itself. In a world of zero-friction copies, the entire premise is bullshit.
I'm disgusted by all the carrying on about copyright. Remember Aaron Swartz
@EUCommission Under no condition. And what was that story about expropropriating EU citizens' data?
@EUCommission
Never.
The LLM are parasitical toys.
Under the same conditions I'm allowed to publish them. WHEN THE RIGHTSHOLDER GRANTED THE RIGHTS?
You know, exactly what is law?
@EUCommission
Only when it is also freely available for humans to train themselves on, i.e. to read.
@a_cubed Absolutely reasonable thought. I would be very happy if you also give this opinion in the survey. Please come in! :)
@mit_scharf @a_cubed I think this was the argument in the US, too, which made it the AI companies had to settle with publishers where they used illegally torrented books to train their AI, but those who put their works available on the open internet, that was not liable for any compensation.
@nicemicro @mit_scharf @a_cubed
Allowing people to read your work is not the same as allowing them to reproduce it and pass it off as their own (whether or not slightly camouflaged), which is what AI is effectively doing.
@KimSJ @mit_scharf @a_cubed I don't agree with the "pass it off as their own" is true here. The AI companies extract the mathematical regularities from whatever they put in the training set, which can be argued to be transformative. Especially that the AI company is not selling images or text, they sell a tool.
You can argue that if the user of an AI prompts it in a way to reproduce something and pass it off as their own, that is on the user, not dissimilar to use a photocopier to copy a book.
@nicemicro @KimSJ @mit_scharf @a_cubed Unless my assumptions about how EU copyright law works are incorrect, the company that makes the software prompted to make works the user passes off as their own should also be liable.
Unlike using other tools to plagiarise works, the contribution to the labour is merely the prompt. I would think a more apt analogy would be commissioning an artist to make something which resembles copyrighted works the client could pass off as their own.
@EUCommission It’s simple: pay me a royalty and a licence fee.
@DaraghOBrien please 🥺 Please give your opinion about this form in the link! I find this attitude you have very important. It should appear in this survey!
@mit_scharf @DaraghOBrien Have you taken a look at the horribly-worded questionnaire? If the object is honest consultation with the masses, then the bureaucrats responsible for the form have failed in their mission.
I don't think the object is honest consultation with the masses.
@SimonRoyHughes @mit_scharf @DaraghOBrien Do you or anyone else have screenshots or the text of the questions? Until Canada actually is an "associate member" (even then idk) I likely don't qualify to fill this out.
@EUCommission you wouldn't download a car...
Install bonfire.cafe
Get the full app experience
@EUCommission I doubt the commission reads replies here, so please also give them your input at the link they posted, under "consultation", https://ec.europa.eu/eusurvey/runner/targetedconsultationcopyright .
Targeted consultation on a potential initiative for a better copyright environment for European creativity and innovation