[ai-control] Re: -04 drafts for evaluation

[Suresh Krishnan <[e-mail]>]

Hi Lila

> On Oct 30, 2025, at 11:45 PM, Lila Bailey <[e-mail]> wrote:
> 
> Thanks Suresh. 
> 
> I understand that the hope is for this vocab to be used with other attachment points, but the only one we know for sure it will be used with is robots.txt, because that's what the attach doc says.

This is not the case. Even if we look solely within the IETF context we are also chartered to do HTTP attachment which is covered by the Content-Usage Header field in the attach document.

> So if it's going to be messy and harmful to end users with the one attachment point we definitely know will be used, then I think we should seriously reconsider this approach.
> 
> [As a side note, we could also reconsider robots.txt as the attachment mechanism, which I have argued for in the past, but I have heard loud and clear from you and others in this group that robots.txt is the most practical way forward because it's the standard IETF controls and it does not control any of the other possible mechanisms that everyone is hoping the vocab will get used with. So I have given up that particular crusade, and am in good faith trying to find a way to make the vocab document work effectively with robots.txt if that's what IETF is going with.]
> 
> In any case, this response doesn't address the core concern, which is that this category is outside of the charter because it goes far beyond AI uses. I would really appreciate hearing how the chairs are justifying going outside the scope of the charter. This is my first IETF rodeo, and I may be missing something here. But I (and many others, as Jo noted) are concerned about this overreach, so I hope that it will be addressed.

If we can gain consensus either with a tighter definition related to AI, or without the category at all that is certainly something we can explore. Please do make text proposals along with your issues, and we can discuss those proposals both on the list and at the WG meetings.

Thanks
Suresh