Re: Next meeting of the TDMRep CG - March 25 , 16.00-17.00 UTC

I forgot to mention a second topic for this meeting: following a discussion at the last group meeting, Sebastian Posth prepared a proposal to add a new "recipe" to TDMRep: a recipe that stores opt-out information in a decentralized registry. 
Sebastian will present his proposal to the group, and we expect feedback from members. 


Best regards,
Laurent Le Meur | EDRLab

> Le 4 mars 2026 à 16:52, Laurent Le Meur <laurent.lemeur@edrlab.org> a écrit :
> 
> Dear TDMRep CG members, 
> 
> Some of us are actively following the IETF AI Pref discussions. A face-to-face meeting for this IETF group is scheduled for April (in Toronto). 
> Before that, it would be interesting to have a TDMRep meeting, and the chairs propose:
> 
> March 25 at 16.00-17.00 UTC, using Google Meet  https://meet.google.com/fwb-hoip-sgr 
> 
> Some thoughts: the IETF current drafts are:
> - Vocabulary: https://datatracker.ietf.org/doc/draft-ietf-aipref-vocab/, with only two terms now: Search and Foundation Model Production. Other terms may be added after the current round of discussion.
> - Attachment: https://datatracker.ietf.org/doc/draft-ietf-aipref-attach/ , which specifies a robots.txt syntax extension. 
> 
> and discussions are still mainly about the definition and scope of "Search".
> 
> Some TDMRep members - those from the book industry especially - are not really bothered by the extent of what "Search" means, because ebooks are not usually crawled. Metadata embedded in EPUB and PDF are the main ways to express an opt-out. These members are especially sensitive to the use of ebooks for RAG (or AI use or Inference), a use case currently moved to the side by the AI Pref group (the reason being a lack of consensus on terms and scope of this practice). 
> 
> Other members, especially from the news industry, still want to opt out of AI-search if they are not happy with how their content is processed. 
> 
> About this topic, please read https://blog.cloudflare.com/uk-google-ai-crawler-policy/
> and the current discussion of the AI Pref discussion around "Clarification on Generative Search and the "Verbatim" Constraint in Section 4.2" at https://mailarchive.ietf.org/arch/browse/ai-control/?. 
> 
> A good topic of discussion for us is therefore: 
> - Do TDMRep CG members have a clear request to make to the IETF AI Pref group, helping them finding a consensus? 
> - If the IETF AI Pref group does not find a consensus in April on the three use cases that appear important to TDMRep folks, i.e., AI training, AI use, and AI search, what do we do as a group? 
>     * Do we adopt the current Cloudflare approach for a TDMRep vocabulary (search, ai-input, ai-train), knowing that Cloudflare is also ready to update their service if the IETF finds a consensus? 
>     * Do we wait for some more time before deciding which way to go? 
> 
> 
> Cordialement, 
> Laurent Le Meur | EDRLab
>