Well thought out and sensible position overall [LWN.net]

Well thought out and sensible position overall

Posted Apr 28, 2026 10:45 UTC (Tue) by bluca (subscriber, #118303)
In reply to: Well thought out and sensible position overall by Karellen
Parent article: The future of AI in Ubuntu

> Copyright should be abolished, but until that happens megacorps should be held to the same standard as regular people.

They are

> What's inconsistent about that?

The inconsistency is that all the copyright reforms to _weaken_ copyright are being vehemently opposed by alleged free software enthusiats who are now arguing for copyright maximalism (the Disney corporation thanks you for your services), instead of being celebrated as they should

to post comments

Well thought out and sensible position overall

Posted Apr 28, 2026 10:51 UTC (Tue) by pizza (subscriber, #46) [Link] (3 responses)

> > Copyright should be abolished, but until that happens megacorps should be held to the same standard as regular people.
> They are

No, they are not.

"When you owe the bank $100, it's your problem. When you owe the bank $100 Million, it's the bank's problem."

> The inconsistency is that all the copyright reforms to _weaken_ copyright are being vehemently opposed by alleged free software enthusiats

That's because this "weakening" only benefits the large-scale takers.

Well thought out and sensible position overall

Posted Apr 28, 2026 11:08 UTC (Tue) by bluca (subscriber, #118303) [Link] (2 responses)

> No, they are not.

Yes, they are

> That's because this "weakening" only benefits the large-scale takers.

No, it benefits everyone. Models can be trained locally with open source software. Thanks to recent laws weakening copyright protection, absolutely everyone is legally allowed to train models on any publicly available dataset, no questions asked. This is an amazing development that furthers the goal of "information yearns to be free", and any self-respecting so-called "free software enthusiast" should be celebrating it day and night. Instead, they are being the MPA's useful idiots and clamoring for copyright maximalism.

Well thought out and sensible position overall

Posted Apr 28, 2026 11:56 UTC (Tue) by pizza (subscriber, #46) [Link] (1 responses)

> Thanks to recent laws weakening copyright protection,

[citation needed] please?

Because IIUC no actual laws have changed; at best there were court cases that upheld already-existing law.

> Thanks to recent laws weakening copyright protection, absolutely everyone is legally allowed to train models on any publicly available dataset, no questions asked

The legal beef revolves around how one *obtains* the training material, not the use of said material for training. [1]

I personally hope Meta prevails here, because "I'm pirating <X> to train my personal AI" will completely eviscerate all copyright(+software) industry enforcement activities. Of course, that won't be allowed to stand, resulting in a veritable popcorn fest of dueling lobbyists, their captured legislators, and ever-more extravagant library+ballroom donations.

(Meanwhile, It's easy to say "information wants to be free" when you're not the one stuck with the hosting bill from everyone's "training" bots)

[1] https://torrentfreak.com/uploading-pirated-books-via-bitt...

Well thought out and sensible position overall

Posted Apr 28, 2026 12:04 UTC (Tue) by bluca (subscriber, #118303) [Link]

> [citation needed] please?

https://eur-lex.europa.eu/eli/dir/2019/790/oj

which is then referenced by

https://eur-lex.europa.eu/eli/reg/2024/1689/oj/eng

> The legal beef revolves around how one *obtains* the training material, not the use of said material for training.

Yes torrenting stuff was just plain stupid of them. The law of course requires that the source material is publicly available, if one has to break the law to obtain it, then no exception applies. We are not there yet - but every step in that direction, however small, is good progress.

Well thought out and sensible position overall

Posted Apr 29, 2026 0:13 UTC (Wed) by cypherpunks2 (guest, #152408) [Link] (1 responses)

> The inconsistency is that all the copyright reforms to _weaken_ copyright are being vehemently opposed by alleged free software enthusiats who are now arguing for copyright maximalism (the Disney corporation thanks you for your services), instead of being celebrated as they should

To be fair, free software _relies_ on copyright. Copyright enforcement is the very reason VMWare got slapped for violating the GPL. Don't get me wrong, I think copyright law in its current state is garbage that benefits the corporations, but FOSS enthusiasts generally realize that copyright, as flawed as it is, is the only tool at their disposal for protecting our freedoms.

I do think there is a moral difference (and there should be a legal difference) between "pirating" something for personal use and abusing someone else's intellectual property for profit. The AI training case is neither. They aren't taking books and selling them for their own profit, nor are they taking them purely to enjoy reading them. They're using it to train an algorithm. I don't think that, alone, is a big deal. There are far worse issues with generative AI, and I fear attacking things from a copyright perspective is going to backfire.

Well thought out and sensible position overall

Posted Apr 29, 2026 8:35 UTC (Wed) by bluca (subscriber, #118303) [Link]

> To be fair, free software _relies_ on copyright.

As a means to an end. The end was, "information yearns to be free". Well, it's a bit more free now, so it's time to rejoice