https://www.dglaw.com/court-rules-ai-training-on-copyrighted-works-is-not-fair-use-what-it-means-for-generative-ai/

It looks like US case law on the subject might only apply for commercial
works where the original copyright holder suffers a loss of earnings
(because capitalism).

But it's an evolving situation without many relevant rulings and could be
decided differently in future or other jurisdictions.



On Wed, 28 May 2025, 08:36 Jonathan Wakely, <cxx_at_[hidden]> wrote:

>
>
> On Wed, 28 May 2025, 00:42 Ville Voutilainen via Std-Proposals, <
> std-proposals_at_[hidden]> wrote:
>
>> On Wed, 28 May 2025 at 02:33, Oliver Hunt via Std-Proposals
>> <std-proposals_at_[hidden]> wrote:
>> >
>> >
>> >
>> > On May 27, 2025, at 7:06 PM, Frederick Virchanza Gotham via
>> Std-Proposals <std-proposals_at_[hidden]> wrote:
>> >
>> > On Mon, May 26, 2025 at 4:48 AM René Ferdinand Rivera Morell wrote:
>> >
>> >
>> > I do not give you permission to use my copyrighted works to train an AI.
>> > And I doubt that ISO would give you permission to use ISO/IEC
>> copyrighted
>> > works to train an AI. If you do, or have, gone down that road.. Remove
>> my
>> > copyrighted content from your AI data immediately.
>> >
>> >
>> >
>> > I haven't read the small print on what you agreed to when you emailed
>> > Nevin to get a document number, but I'll give consideration to it
>> > before releasing version 1 of my program. I think you retain the
>> > copyright of what you wrote, but obviously if you submit a paper,
>> > other people have to be able to work with it. I can't just write a
>> > paper and submit it and mandate that everyone has to be drinking
>> > decaffeinated coffee when they read it, or that they can store it on
>> > an USB stick but only if the USB stick was purchased on a night when
>> > there was a full moon.
>> >
>> > Well now you have one person saying that they do not give you
>> permission to use their work.
>> >
>> > For you to copy people’s work you need to have the right to do so, I’m
>> not sure what the exact rules for ISO are, but I can guarantee they aren’t
>> “papers are in the public domain” which means you need the permission of
>> the copyright owners. I’m not sure if that’s the authors or ISO, but it
>> sounds like you haven’t even tried to find out, and instead were planning
>> on copying papers without the consent of the authors or the copyright
>> owners (which may or may not be different[1]), and then claiming that your
>> “AI” is something other than a copy of other people’s work.
>> >
>> > Even if we were to try to assume goodwill on your part, when someone
>> said you did not have their consent to steal their work, you said “I don’t
>> care, I’m going to do it anyway”. That you think that is a reasonable, or
>> even remotely ethical, behavior is absurd. Step 1 of being in any community
>> is demonstrating some amount of respect for others in the community, and
>> you haven’t demonstrated even the slightest semblance of that.
>> >
>> > Just to be clear, you also don’t have my consent to steal any of my
>> work either.
>> >
>> > —Oliver
>> >
>> >
>> >
>> > [1] Per https://www.iso.org/copyright.html all ISO documentation is
>> copyright protected, whether ISO has copyright ownership of a proposal, or
>> if the author retains it (I’m unclear whether a proposal is considered an
>> ISO publication), is moot: it’s covered by copyright either ISO or the
>> authors, and so copying the work requires permission of the copyright owner.
>>
>> The proposals aren't ISO publications, they are not in any ISO system
>> (that's why they are P-numbered, not N-numbered), so the only possible
>> copyright holders
>> for them are their authors.
>>
>> IANAL, of course. Don't take that as actual legal advice. It's my
>> understanding of the situation where there is no copyright assignment
>> being made,
>> and no distribution or other kind of license being stated.
>>
>> The copyright of the proposed wording transfers to ISO when it's
>> incorporated into the draft and subsequently to a standard , and
>> proposal authors understand that,
>> but they still hold the sole copyright to their proposal papers,
>> because nothing else is stated.
>>
>
> Right, authors retain copyright of papers, and as Oliver said, they're
> certainly not in the public domain. That's a specific legal term that means
> they're *not* covered by copyright, so you can't say they're copyrighted
> and also in the public domain, that's a contradiction.
>
> The permission to redistribute the papers (which is needed for them to be
> hosted online and sent out to the committee) does not imply permission to
> modify them and create derived works (such as a training set for an LLM). I
> believe you would need permission from the individual authors for that.
>
> To be clear, I don't think you have any implied permission, the default is
> no permission, and you would have to obtain it from every author. (For the
> avoidance of doubt, I don't give permission for any of my papers to be used
> for training LLMs.)
>
> I don't know what the legal situation is if you never share the work based
> on that training set, and it's only for your personal use on your own
> computer. It might be that we can't prevent you from doing that. But
> hosting the service online for others to use would be distributing derived
> works without permission.
>
>
>
>