Re: AI slop "contributions" to IETF working groups

["Rob Wilton (rwilton)" <[e-mail]>]

I sort of agree with Chris here.

AI is a tool that can improve productivity.  We should allow, probably encourage, usage of tools that increase productivity where possible.  One of the drafts (for guidelines not a protocol) that I have recently submitted was predominately written by AI and it has done a pretty good job (probably much better than my first attempt would have been, and significantly quicker).  Yes, I obviously reviewed and improved the content in various places before submitting, and yes it still requires more work.  When I posted the initial draft I made it very clear to those that I was asking for reviews that it was mostly AI generated.

Yes, AI can also produce garbage, just like humans can, and I completely agree that just spamming rubbish material to an alias or WG is not good behaviour, it just wastes time.

However, whether this will really be a problem is unclear.  Part of the IETF motto is "We reject: kings, presidents, voting."  But instead, I perceive that IETF is built on technical excellence and mutual respect.  E.g., if a respected participant starts sending lots of rubbish AI generated messages or documents then they will quickly lose the respect of their peers because they are wasting everyone's time.

I think that we should acknowledge in documents when AI has been used to write a significant amount of the content.  Beyond that, I'm not sure that I see any great value in a flag to indicate that it has been reviewed by a human before sending, since every email that I send to the list or draft that I post with my name as an author implies that I am willing to stand by the content within.

Kind regards,
Rob


From: Christian Hopps <[e-mail]>
Date: Tuesday, 10 February 2026 at 11:37
To: Brian E Carpenter <[e-mail]>
Cc: [e-mail] <[e-mail]>
Subject: Re: AI slop "contributions" to IETF working groups

Brian E Carpenter <[e-mail]> writes:

> On 08-Feb-26 14:55, John Levine wrote:
>> It appears that Bob Hinden  <[e-mail]> said:
>>> I like the idea of requiring a disclosure that an Internet Draft was written with AI tools.  I agree probably not for emails.
>> I expect that the people whose drafts most suffer from being written by an LLM
>> will lie about it, either because
>> their English is too poor to understand the rules, or they have perverse incentives like an an employer that
>> gives them a bonus for publishing an RFC.
>> Then what?  I’m not saying we shouldn’t ask, but I’m saying we need to be
>> prepared for noncompliance.
>
> Of course, but if we make it a submission requirement - check the “AI used” box - there’s really no harm, no shame.
>
> Actually there are perhaps two boxes to check:
>
> 1. Was AI used in the production of this draft? [yes/no]
> 2. If yes, did the authors personally verify the AI-generated material? [yes/no]

So we’ve been discussing this recently in the FRR development community, and we decided sort of the opposite direction. We require a review and approval for all code pull-requests before they can be merged. Since a code submission will either be good and acceptable, or it won’t be, the fact that AI tools were used to a greater or lesser or no extent really doesn’t matter and we didn’t think it should affect the accept/reject decision.

So anyway, while we’re (IETF) considering requiring disclosure of AI tool use here, I think it’s worth considering what exactly we’d like this disclosure to accomplish. Is it a filter flag (i.e., if checked “Yes” it get’s dropped to /dev/null by a personal email filter)? Does it help in reviewing the document knowing that AI tools were used.

I *think* what’s being looked for here really is a filter-flag to allow people to ignore submissions with the “wrong” disclosure (e.g., “AI was used, Author did not review” :) In this case everyone is also going to realize that that’s what it’s being used for. Given that, do we really think someone who is willing to post a document as their own work, but also not review that document themselves, is really going to also be honest and admit that? I have my doubts.

Thanks,
Chris.


> The attachment was generated by ChatGPT.
>
>    Brian
>
> [2. image/svg+xml; qq.svg]…