[ai-control] Re: Proposed "AI System Inference" and "AI User Input" categories

[Jo Levy <[e-mail]>]

I agree with Farz that the difficulty with AI inference is that the technology, and the common understanding of the term, are rapidly evolving.


From: Leonard Rosenthol <lrosenth=[e-mail]>
Sent: Friday, July 31, 2026 12:36 PM
To: Farzaneh Badiei <[e-mail]>; Laurent Le Meur <[e-mail]>
Cc: [e-mail]
Subject: [ai-control] Re: Proposed "AI System Inference" and "AI User Input" categories

Farz - completely understand your position on this, but it is just one voice among many - and there are voices on the other side that feel strongly about the need for this option.  I would expect/hope that, as we have done in the past, we will spend some time at our F2F in London next month on this topic.  Clearly there is interest in bringing it back, now that we have 3.2 to back it up.

Leonard

From: Farzaneh Badiei <[e-mail]<mailto:[e-mail]>>
Date: Friday, July 31, 2026 at 2:18 PM
To: Laurent Le Meur <[e-mail]<mailto:[e-mail]>>
Cc: [e-mail]<mailto:[e-mail]> <[e-mail]<mailto:[e-mail]>>
Subject: [ai-control] Re: Proposed "AI System Inference" and "AI User Input" categories

EXTERNAL: Use caution when clicking on links or opening attachments.


Hi Laurent,

Yes this topic keeps being brought up by publishers and we continue expressing concerns about including inference at this stage. we didn't have consensus on including this category, and there are real open questions about scope and definitions that haven't been resolved. I am not saying we should never include it. We can look into it later, once we better understand the impact and the technology has stabilized enough that it won't quickly change its meaning. In my opinion, inference is not mature enough to be standardized. We should not define it based on our own assumptions. This needs to be discussed with the broader group working on inference today, not settled here based on our current understanding of its meaning. Maybe we can start an IRTF RG to discuss this more or have a group under one of the RGs to discuss. I am not so sure how that process works so I might be wrong.

We are doing this at the IETF, it is a global standard process. RSL isn't a formal open standards body like the IETF or even C2PA. Both C2PA and RSL also have problems and risks regarding copyright overreach and the impact on consumer and end user access doesn't disappear just because others are discussing it.

I have always mentioned that 3.2 is good to have but it is not enough.If we use asset-level signaling, client system developers may decide not to allow their users to upload simple PDFs based on these signals. We have mentioned the problems this could raise. It is not an effective way to prevent undesirable content, but it can significantly hamper consumers' ability to use translation services, for example.
Another group of rights-owners in this equation includes end users and humanrights-holders. Many times the adverse effect of copyright overreach goes unnoticed and gets lost when the focus is solely on meeting the needs of copyright holders.

On Thu, Jul 30, 2026 at 10:10 AM Laurent Le Meur <[e-mail]<mailto:[e-mail]>> wrote:
Hi Farzaneh,

Rights owners have expressed their need for a preference regarding AI inference opt-out since the start of this IETF project, and even before. Such preference/control is already present in multiple opt-out models (C2PA, RSL ...).

Section 3.2 covers the fact that AI preferences may not be followed by users and client systems. Therefore, the specification leaves the evaluation of their impact on accessibility, translation, and other user-beneficial features to the client systems and their users. Not us.
Thanks to section 3.2, the document you reference is now mainly of interest to developers of client systems.

Best regards
Laurent LE MEUR


Le 30 juil. 2026 à 15:02, Farzaneh Badiei <[e-mail]<mailto:[e-mail]>> a écrit :

One thing that warrants a long and in-depth discussion is that since the beginning of this working group, some of us time and time again have warned against having an inference category especially for the consumer. It can have an impact the use of accessibility and translation features along with many more issues. Meredith, Lila, Jo and myself have time and time again raised these concerns in different shape and form. I raised them last week as well.

We had submitted this draft to illustrate some of the impact. https://www.ietf.org/archive/id/draft-farzdusa-aipref-enduser-00.html

In my opinion, we need to take a step back and before we discuss any text, we should discuss the addition of inference, whether the group agrees with it and under what circumstances the group agrees with adding inference.