Re: AI disclosure [was: AI slop "contributions" to IETF working groups]

[Abdussalam Baryun <[e-mail]>]

Hi Philip,

On Sat, Feb 14, 2026 at 5:18 AM Phillip Hallam-Baker <[e-mail]>
wrote:

>
>
> On Fri, Feb 13, 2026 at 6:23 PM Christian Huitema <[e-mail]>
> wrote:
>
>>
>> On 2/12/2026 5:56 PM, Phillip Hallam-Baker wrote:
>> > On Wed, Feb 11, 2026 at 9:36 AM Michael Richardson <
>> [e-mail]>
>> > wrote:
>> >
>> >> John Levine <[e-mail]> wrote:
>> >>      > It appears that Brian E Carpenter <[e-mail]>
>> >> said:
>> >>
>> >>      > In the United States, the answer is nobody.  US courts have
>> held that
>> >>      > only people can create copyrighted or patented material, not
>> robots
>> >> or
>> >>      > monkeys.*
>> >>
>> >> So, if IETF (Trust) can't own/license a copyright on an I-D/RFC
>> produced
>> >> like
>> >> that, then I don't think it can pass NoteWell.
>> >>
>> > I am not a lawyer but, I do have frequent discussions with lawyers on
>> such
>> > questions.
>> >
>> > The Monkey copyright case is not quite as simple as 'can't copyright
>> > anything not made by a human'.
>> >
>> > The place where I expect the courts to eventually end up is to ask, 'was
>> > the prompt that generated the output copyrightable'. If so, the output
>> is
>> > copyrightable.
>> >
>> > That might not be the first ruling that comes out but it is the only
>> > approach I can see being sustainable.
>>
>> That's your opinion, but I think reducing that to "the prompt" is a
>> snapshot of the text generating
>> LLM technology as of 2025. We can find lots of precedents elsewhere. For
>> example, is Duchamp's
>> "Fountain" copyrightable? I would not expect a one size fits all answer.
>> What if many paragraphs of
>> the text come from AI and many don't, or they were partly rewritten, or
>> they were rearranged?
>> The nuanced answer has the property of requiring lots of lawyer time to
>> sort out nuances, and
>> thus is likely to be favored by the legal profession.
>>
>
> I was only looking at the very narrow question of whether the user of an
> LLM can have a copyright interest in the output of an LLM. It is pretty
> clear from similar cases that they can. It is also pretty clear that in
> many cases, they cannot establish an exclusive copyright interest due to
> the widespread use of copyright materials in building the models in the
> first place.
>

I think because of AI, the international copyright rules will change in
future, hopefully to assist human rights to use technology. Some thing that
current copyright rules are not helping human rights, so that needs more
discussions in non_technology_SDOs.


> Which bringing it back to Internet Drafts, might well mean that we can't
> accept AI slop because the prompt author has no idea whether they are
> infringing.
>
> The merchants of slop might be able to buy themselves a copyright act that
> absolves them and their products from liability in the US but that isn't
> going to work for us because we publish internationally.
>

I agree, that is why IETF_discussions/meetings are the key for best
practices/human_rights. The human rights are very important to be redefined
after the development of AI technologies, and we should not let machines or
engineers become against human rights.

AB

AB