Re: Dealing with LLMs in IETF discussions draft

This thread looks more an more like i imagine the nightmare version of
the early years of cars: Let's just legislate on how to best hang people who
cause problems with these new dangerous toys - but lets not invest into
how to make cars and roads better and safer or educate drivers.

Please, someone, anyone, Buehler ? What's needed is guidance on how to
best use/benefit from AI in the IETF (discussions). More of that, less of this here.

Cheers
    Toerless

On Wed, Aug 05, 2026 at 01:54:44PM +1000, George Michaelson wrote:
> I think in strict legalisms I agree, IFF we impose the barrier to insist
> the human review happens, and we insist that "you posted it you said it"
> even if an AI wrote it to be cut-and-pasted. That insistance is a process
> cost we have to implement and review, in my opinion.
> 
> I don't think we can just sit back and assume we're ok. I think we should
> take an active position in this and declare requirements which suit our
> needs.
> 
> I didn't post this to achieve some mythical strawman slam-dunk outcome. I
> think it says things about what legislature thinks about AI, and what kinds
> of things we might want to think about, using AI. Not thinking about it is
> also not in my goal set of course: I think we should have a conscious
> documented path here.
> 
> -G
> 
> On Wed, Aug 5, 2026 at 1:51 PM Brian E Carpenter <
> [email protected]> wrote:
> 
> > On 05-Aug-26 13:27, George Michaelson wrote:
> > > They provide an explorer function over the legislation:
> > >
> > > https://ai-act-service-desk.ec.europa.eu/en/ai-act-explorer <
> > https://ai-act-service-desk.ec.europa.eu/en/ai-act-explorer>
> > >
> > > I hope that helps.
> >
> > Yes, it does, thanks. In clause 4 of [1] we find:
> >
> > "Deployers of an AI system that generates or manipulates text which is
> > published with the purpose of informing the public on matters of public
> > interest shall disclose that the text has been artificially generated or
> > manipulated. This obligation shall not apply where the use is authorised by
> > law to detect, prevent, investigate or prosecute criminal offences or where
> > the AI-generated content has undergone a process of human review or
> > editorial control and where a natural or legal person holds editorial
> > responsibility for the publication of the content."
> >
> > Human review and responsibility rule, OK? So I don't think this EU law has
> > any impact on the text of RFCs, even assuming that RFCs cover "matters of
> > public interest".
> >
> > [1] https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-50
> >
> >      Brian
> >

-- 
---
[email protected]