https://notes.ietf.org/notes-ietf-interim-2025-aipref-08-aipref
", "time": "2025-06-23T13:01:08Z"}, {"author": "Martin Thomson", "text": "Oslo is where it's all happening
", "time": "2025-06-23T13:04:13Z"}, {"author": "Martin Thomson", "text": "Makes you want to take the direct
", "time": "2025-06-23T13:04:38Z"}, {"author": "Timid Zehta", "text": "Greetings, I'm Timid Robot (they/them). I reside in Malta and work with Creative Commons.
", "time": "2025-06-23T13:06:40Z"}, {"author": "Thom Vaughan", "text": "Hi y'all, nice to meet/see you. I'm Thom, and I work at Common Crawl.
", "time": "2025-06-23T13:07:30Z"}, {"author": "Brendan Quinn", "text": "Hi everyone. Sorry but I won't be very active in the meeting, I'm in a house full of screaming kids (+ a few adults) right now because it's basically the equivalent of Estonian thanksgiving today and all the extended family is together at our summer house. But I'll contribute on chat where I can!
", "time": "2025-06-23T13:09:29Z"}, {"author": "Brendan Quinn", "text": "(I work for IPTC which is based in the UK but I live in Estonia)
", "time": "2025-06-23T13:09:52Z"}, {"author": "Suresh Krishnan", "text": "No worries Brendan. Totally get it. Please feel free to participate in chat to ask questions or provide comments.
", "time": "2025-06-23T13:13:13Z"}, {"author": "Farzaneh", "text": "did we define TDM so narrowly so that we can can say no to TDM and yes to AI? (I have been following the definition changes and I am not sure)
", "time": "2025-06-23T13:13:33Z"}, {"author": "St\u00e9phane Bortzmeyer", "text": "From figure 1 in the draft, AI is included in TDM
", "time": "2025-06-23T13:14:08Z"}, {"author": "Timid Zehta", "text": "\"tdm=n,ai=y\" is a valid expression
", "time": "2025-06-23T13:14:43Z"}, {"author": "Farzaneh", "text": "yes but if you say no to TDM can you say yes to AI?
", "time": "2025-06-23T13:14:49Z"}, {"author": "Timid Zehta", "text": "yes
", "time": "2025-06-23T13:14:57Z"}, {"author": "Bryan Newbold", "text": "@Farzaneh I believe the intent of a preference like \"tdm=n, ai=y\" would indicate that AI (the more specific case) is \"yes\", but any other non-AI TDM uses would be \"no\"
", "time": "2025-06-23T13:15:06Z"}, {"author": "Paul Keller", "text": "yes
", "time": "2025-06-23T13:15:06Z"}, {"author": "St\u00e9phane Bortzmeyer", "text": "Then figure 1 of the draft is wrong.
", "time": "2025-06-23T13:15:08Z"}, {"author": "Farzaneh", "text": "yes Stephane I think we need to do a thorough edit of that figure
", "time": "2025-06-23T13:15:36Z"}, {"author": "Gary Illyes", "text": "PETA FTW
", "time": "2025-06-23T13:15:49Z"}, {"author": "Paul Keller", "text": "no it is not wrong is just shows the hierarchy. this needs to be read in combination with the rule that a more specific statement overrules a less specific one
", "time": "2025-06-23T13:15:55Z"}, {"author": "Farzaneh", "text": "it doesnt show the hierarchy well
", "time": "2025-06-23T13:16:07Z"}, {"author": "Paul Keller", "text": "it will be improved
", "time": "2025-06-23T13:16:19Z"}, {"author": "Suresh Krishnan", "text": "@St\u00e9phane and @Farz the current updated figure in the editor's copy of the draft is at https://ietf-wg-aipref.github.io/drafts/draft-ietf-aipref-vocab.html#f-categories
", "time": "2025-06-23T13:18:38Z"}, {"author": "Farzaneh", "text": "I have concerns about RAG and Inference and Search. will send my comments to the mailing list.
", "time": "2025-06-23T13:18:44Z"}, {"author": "Suresh Krishnan", "text": "Sounds good Farz.
", "time": "2025-06-23T13:19:15Z"}, {"author": "Brendan Quinn", "text": "I'm happy to change the names of categories to make them more neutral, but if we do, can we be clear about how our definitions relate to the ones in the EU Act (and maybe others?)? Is there precedent for doing that (referring to specific legislation) in RFCs?
", "time": "2025-06-23T13:19:28Z"}, {"author": "Farzaneh", "text": "thanks Suresh. that figure doesn't show hierarchy. it kind of implies that TDM includes all of these functions.
", "time": "2025-06-23T13:19:30Z"}, {"author": "Martin Thomson", "text": "If you have anything that you can briefly help us with today Farzaneh, that will be helpful.
", "time": "2025-06-23T13:19:30Z"}, {"author": "Farzaneh", "text": "I also have concerns about using exact phrases that the EU uses in its legislation. sorry yeah I will try to be more constructive... maybe during this week and not today
", "time": "2025-06-23T13:20:52Z"}, {"author": "Leonard Rosenthol", "text": "@brendan - that's been part of the issue is that we don't want to use \"loaded terms\" (like TDM) so that people don't read things as directly mapping to any legislatiion.
\nSuch mappings need to take place in other documents.
", "time": "2025-06-23T13:20:53Z"}, {"author": "Paul Keller", "text": "@farzaneh But that is hierarchy: all the other categories are subsets of TDM
", "time": "2025-06-23T13:21:01Z"}, {"author": "Brendan Quinn", "text": "@leonard okay, which documents should those be and how can we help to write them? :-)
", "time": "2025-06-23T13:21:56Z"}, {"author": "Martin Thomson", "text": "Please don't fixate too much on the figure. The words are what matters.
", "time": "2025-06-23T13:22:23Z"}, {"author": "Leonard Rosenthol", "text": "@brendan, that isn't clear right and also where those docs will come from (IETF, etc.)
", "time": "2025-06-23T13:22:35Z"}, {"author": "Martin Thomson", "text": "TDM is defined in a broad way; search and inference are categories of use that we have discussed being a lot more narrow.
", "time": "2025-06-23T13:22:56Z"}, {"author": "Brendan Quinn", "text": "+1 I would vote for search being outside of the box
", "time": "2025-06-23T13:25:39Z"}, {"author": "Timid Zehta", "text": "the legal implications of the preferences will depend, in part, on whether the declaring party is the rights holder or represents the rights holder.
\nthe ai preferences, themselves, don't necessarily need to know/indicate whether the declaring party is operating under copyright law or not
", "time": "2025-06-23T13:25:46Z"}, {"author": "Farzaneh", "text": "we have to be clear about that
", "time": "2025-06-23T13:26:13Z"}, {"author": "Timid Zehta", "text": "@Farzaneh please make a case for including copyright status in ai preferences.
", "time": "2025-06-23T13:27:44Z"}, {"author": "Farzaneh", "text": "if AI inference is used in training how can we say no to AI training and yes to AI inference for example? it's just not that clear cut in my opinion.
", "time": "2025-06-23T13:27:47Z"}, {"author": "Leonard Rosenthol", "text": "AI inference is NOT used in AI training - they are completely separate.
", "time": "2025-06-23T13:28:52Z"}, {"author": "Martin Thomson", "text": "Farz, the point I make here is that there is an application with a purpose and the \"AI Training\" purpose subsumes the \"AI Inference\" that might be used in that training.
", "time": "2025-06-23T13:29:03Z"}, {"author": "Farzaneh", "text": "yeah I have some suggestions for that language
", "time": "2025-06-23T13:29:14Z"}, {"author": "Martin Thomson", "text": "Leonard, inference can be used in training, sadly. According to some articles, DeepSeek was trained on generated content from OpenAI models. More generally, models are being trained on generated content (to their detriment, you might say, but this is part of the science of training).
", "time": "2025-06-23T13:30:08Z"}, {"author": "Brendan Quinn", "text": "@martin but we define inference as \"The act of using one or more assets as input to a trained AI/ML model as part of the operation of that model (as opposed to the training of the model).\" So by definition they are different...?
", "time": "2025-06-23T13:30:59Z"}, {"author": "Farzaneh", "text": "Thank you Paul. I will make some hopefully helpful suggestions on the list :)
", "time": "2025-06-23T13:31:04Z"}, {"author": "Timid Zehta", "text": "i see significant value in having a super category (ex. TDM)
", "time": "2025-06-23T13:32:06Z"}, {"author": "Leonard Rosenthol", "text": "Correct Brendan!
\nThe most common use case for inference is \"style reference\" - where a user \"uploads\" an asset to an AI system to accompany a prompt that said \"make it like this one\"
", "time": "2025-06-23T13:32:13Z"}, {"author": "Brendan Quinn", "text": "from a usage point of view, I could imagine people allowing usage for search but not for any type of TDM (or whatever we call it). So from that point of view it would be helpful for them to be different.
", "time": "2025-06-23T13:32:32Z"}, {"author": "Martin Thomson", "text": "Brendan: good point, I had missed that parenthetical. It seems we already solved that problem.
", "time": "2025-06-23T13:33:15Z"}, {"author": "Farzaneh", "text": "Super category that can potentially include all the functions and you can say no to? what are these types of TDMs?
", "time": "2025-06-23T13:33:19Z"}, {"author": "Mirja K\u00fchlewind", "text": "I'm not sure if there is a good use case for having a hierarchy. A default catch-all might not always be what you really want for future use cases.
", "time": "2025-06-23T13:34:19Z"}, {"author": "Paul Keller", "text": "if we really want to narrow this down we need to define what \"AI\" is, but I would suggest that we do not open that can of worms
", "time": "2025-06-23T13:34:29Z"}, {"author": "Brendan Quinn", "text": "without a hierarchy, we need to opt out of every usage separately which becomes a usage nigtmare.
", "time": "2025-06-23T13:34:54Z"}, {"author": "Timid Zehta", "text": "+1
", "time": "2025-06-23T13:35:11Z"}, {"author": "Mark Nottingham", "text": "Starting to wonder if TDM is just \"non-human usage\"
", "time": "2025-06-23T13:35:22Z"}, {"author": "Farzaneh", "text": "but you do that already by having a super broad category
", "time": "2025-06-23T13:35:59Z"}, {"author": "Mark Nottingham", "text": "yes
", "time": "2025-06-23T13:36:16Z"}, {"author": "Martin Thomson", "text": "Yes, TDM is a super-broad category. That's also true of the definitions in EU law, for instance.
", "time": "2025-06-23T13:36:54Z"}, {"author": "Leonard Rosenthol", "text": "@mark - yes, pretty much.
\nI found this definition for it earlier today: \"deriving information from machine-read material\"
", "time": "2025-06-23T13:37:03Z"}, {"author": "Martin Thomson", "text": "Or even \"computers processing stuff\".
", "time": "2025-06-23T13:37:22Z"}, {"author": "Mark Nottingham", "text": "but its not that simple. My web browser is a computer processing stuff. How far from the user do you have to be? Some amount of vagueness is OK, but we need some guidance to align expectations.
", "time": "2025-06-23T13:38:40Z"}, {"author": "Martin Thomson", "text": "Ste'phane: yes, that is exactly the point of the broad category.
", "time": "2025-06-23T13:39:03Z"}, {"author": "Farzaneh", "text": "+1 St\u00e9phane.
", "time": "2025-06-23T13:39:10Z"}, {"author": "Mark Nottingham", "text": "We seem to be seeing a bit of a disconnect about the role of hierarchy.
", "time": "2025-06-23T13:39:37Z"}, {"author": "Brendan Quinn", "text": "I would say difference in understanding rather than difference in principles!
", "time": "2025-06-23T13:39:59Z"}, {"author": "St\u00e9phane Bortzmeyer", "text": "@Martin : so if you say tdm=n, you cannot say yes to others?
", "time": "2025-06-23T13:40:04Z"}, {"author": "Bryan Newbold", "text": "I assume we don't want to include simple transport of data in the enclosing category. And probably not some categories of processing like transcoding image/video, or converting from one encoding to another.
\nMaybe \"non-consumptive\" gets at the semantics?
you can say \"tdm=n,search=y\", for example
", "time": "2025-06-23T13:40:28Z"}, {"author": "Martin Thomson", "text": "@St\u00e9phane: the current structure allows you to say tdm=n, search=y, which specifically allows search.
", "time": "2025-06-23T13:40:34Z"}, {"author": "Mirja K\u00fchlewind", "text": "why is that better than just \"search=y\"?
", "time": "2025-06-23T13:40:46Z"}, {"author": "St\u00e9phane Bortzmeyer", "text": "@Timid then it would defeat the entire point of having a hirearchy
", "time": "2025-06-23T13:40:47Z"}, {"author": "Farzaneh", "text": "but Martin we are not clear what constitutes TDM. also I thought we dont want to use the term TDM but just define it
", "time": "2025-06-23T13:40:57Z"}, {"author": "Martin Thomson", "text": "Mirja, because it excludes unknown future categories of use that might be defined under TDM.
", "time": "2025-06-23T13:41:07Z"}, {"author": "Martin Thomson", "text": "TDM is very clearly defined (in my view)
", "time": "2025-06-23T13:41:19Z"}, {"author": "Mirja K\u00fchlewind", "text": "\"search=y\" excludes everything else (if we define it that way)
", "time": "2025-06-23T13:41:36Z"}, {"author": "Brendan Quinn", "text": "@stephane I disagree, that's exactly why we need a hierarchy. Otherwise you must list tdm=n,genai=n,aiinference=n,...,search=y
", "time": "2025-06-23T13:41:42Z"}, {"author": "Mark Nottingham", "text": "search=y only speaks about search
", "time": "2025-06-23T13:41:51Z"}, {"author": "Martin Thomson", "text": "Mirja: it does not: saying nothing about something says nothing about it, not positive, not negative
", "time": "2025-06-23T13:42:05Z"}, {"author": "St\u00e9phane Bortzmeyer", "text": "@Brendan No, since they are all par of TDM, no need to specify them
", "time": "2025-06-23T13:42:09Z"}, {"author": "Mark Nottingham", "text": "maybe tdm= is really just default=
", "time": "2025-06-23T13:42:15Z"}, {"author": "Mark Nottingham", "text": "ie it's the only one with special status
", "time": "2025-06-23T13:42:30Z"}, {"author": "Leonard Rosenthol", "text": "but we agreed that there is no \"default\"
", "time": "2025-06-23T13:42:32Z"}, {"author": "Mirja K\u00fchlewind", "text": "how do I say then that I only want to allow search and nothing else? Maybe there is something else than TDM in future?
", "time": "2025-06-23T13:42:43Z"}, {"author": "Brendan Quinn", "text": "@stephane but you just defined a hierarchy :-)
", "time": "2025-06-23T13:42:45Z"}, {"author": "Martin Thomson", "text": "I did consider proposing \"tdm\" be renamed to \"all\".
", "time": "2025-06-23T13:42:46Z"}, {"author": "Mirja K\u00fchlewind", "text": "and I'm saying we don't need an explicit \"all\"
", "time": "2025-06-23T13:43:07Z"}, {"author": "Paul Keller", "text": "Mirja you say search=y
", "time": "2025-06-23T13:43:24Z"}, {"author": "Felix Reda", "text": "I can't take consistent notes of the chat discussion, btw. Feel free to also add them to the notes doc, if relevant.
", "time": "2025-06-23T13:43:34Z"}, {"author": "Paul Keller", "text": "and tdm=n
", "time": "2025-06-23T13:43:35Z"}, {"author": "Martin Thomson", "text": "Mirja: I firmly disagree. If you say search=y, that does not imply anything about uses outside of search.
", "time": "2025-06-23T13:43:41Z"}, {"author": "Mark Nottingham", "text": "That's fine Felix.
", "time": "2025-06-23T13:43:47Z"}, {"author": "Leonard Rosenthol", "text": "+1 to Martin
", "time": "2025-06-23T13:43:50Z"}, {"author": "Paul Keller", "text": "sorry I meant to say search=y and tdm=n
", "time": "2025-06-23T13:44:03Z"}, {"author": "St\u00e9phane Bortzmeyer", "text": "Bikeshedding: I like NHU (Non-Human Use) instead of TDM or ALL.
", "time": "2025-06-23T13:44:32Z"}, {"author": "Brendan Quinn", "text": "I can't put my hand up due to background noise, but I think we could help discussions through use cases / examples. It might help to clarify what we're actually trying to achieve and how each example can be implemented using our various draft syntax approaches.
", "time": "2025-06-23T13:44:49Z"}, {"author": "Brendan Quinn", "text": "I tried to make a start on some use cases but didn't have time to bring it to a useful state before this call.
", "time": "2025-06-23T13:45:23Z"}, {"author": "Glenn Deen", "text": "could there be a high level assertion to say \"my following answers are using opt-in\" or conversely \"opt-out\". which declares the intent
", "time": "2025-06-23T13:47:19Z"}, {"author": "Mirja K\u00fchlewind", "text": "but that's the point. we don't have to define the regime. we should leave that to the context that it is used in.
", "time": "2025-06-23T13:48:02Z"}, {"author": "Brendan Quinn", "text": "TBH I never liked \"inference\" as a name either :-) maybe \"at prompt time\" vs \"at training time\" or something else.
", "time": "2025-06-23T13:49:27Z"}, {"author": "Brendan Quinn", "text": "@felix surely the rights holder of the media has a say in how their creative work is used at \"inference time\" (or whatever we call it?)
", "time": "2025-06-23T13:49:51Z"}, {"author": "Farzaneh", "text": "thats an interesting suggestion Brendan
", "time": "2025-06-23T13:49:59Z"}, {"author": "Bradley Silver", "text": "To Felix's point about putting content into a prompt, is this not in fact - practically - invisible and outside the scope of what we are doing here? How would a preference be attached to that, even?
", "time": "2025-06-23T13:50:16Z"}, {"author": "Brendan Quinn", "text": "embedded metadata
", "time": "2025-06-23T13:50:29Z"}, {"author": "Brendan Quinn", "text": "which I wanted to bring up :-)
", "time": "2025-06-23T13:50:37Z"}, {"author": "Suresh Krishnan", "text": "@Mirja, agree we don't have to define the regime but having a catchall bucket allows a person making a statement of preference to have consistent behavior under different regimes.
", "time": "2025-06-23T13:50:39Z"}, {"author": "Mirja K\u00fchlewind", "text": "we could define a catchall if needed but it doesn't have to be a hierarchy. if that is a separate function it would give more flexibility to use it or not.
", "time": "2025-06-23T13:51:38Z"}, {"author": "Farzaneh", "text": "Perhaps Bradley, but should they assert that right through robots.txt and this vocab? also as I said we keep talking about this as if all the declarants are rightsholders. they arent
", "time": "2025-06-23T13:51:58Z"}, {"author": "Jessie Stricchiola", "text": "@leonard - during RAG, reranker models are trained on the data obtained during retrieval. So technically information from retrieved documents used at \"inference\" is involved in subsequent training of these systems.
", "time": "2025-06-23T13:52:33Z"}, {"author": "Paul Keller", "text": "in that case the question if you can use that (you someone running such a system) depends not only on the inference/RAG preference but also on the AI training preference. In the end it is up to the user to interpret the preferences
", "time": "2025-06-23T13:53:44Z"}, {"author": "Leonard Rosenthol", "text": "@jessie understood, and that is clearly another good reason why we should split up RAG from the \"prompt inference\"
", "time": "2025-06-23T13:55:07Z"}, {"author": "Farzaneh", "text": "+1 Felix
", "time": "2025-06-23T13:57:34Z"}, {"author": "Timid Zehta", "text": "still a UX issue. the application could include different capabilities for different users
", "time": "2025-06-23T13:57:41Z"}, {"author": "Glenn Deen", "text": "+! Felix
", "time": "2025-06-23T13:57:45Z"}, {"author": "Aditya Kanteti", "text": "Here is a definition of TDM I found online:
\n\"Text and Data Mining (TDM) is the process of deriving information from machine-read material. It works by copying large quantities of material, extracting the data, and recombining it to identify patterns\"
\n-www.libereurope.eu
_Consumed_ but not crawled
", "time": "2025-06-23T14:01:28Z"}, {"author": "Thom Vaughan", "text": "Thank you for that distinction
", "time": "2025-06-23T14:01:48Z"}, {"author": "Mark Nottingham", "text": "Inference doesn't fit into that definition (as discussed)
", "time": "2025-06-23T14:01:58Z"}, {"author": "Brendan Quinn", "text": "I need to jump but thanks for the great discussion. I look forward to continuing in London.
\nOne point I'd like to raise in the attachment topic is to allow for embedded metadata. We at IPTC are in principle willing to align our existing embedded metadata data mining vocabulary with the vocabulary defined in this group. But it would be great if we could allow for metadata embedded directly in assets to first-level consideration in the spec! I think Leonard is aligned on this point (but please say so if not, Leonard!).
", "time": "2025-06-23T14:02:10Z"}, {"author": "Martin Thomson", "text": "Brendan: we have that, we will have to rely on people like yourself to define those.
", "time": "2025-06-23T14:02:45Z"}, {"author": "Leonard Rosenthol", "text": "@brendan - yes, attachment absolutely includes embedded metadata (IPTC, C2PA, etc.)
", "time": "2025-06-23T14:02:48Z"}, {"author": "Martin Thomson", "text": "{ \"ai\": true } would be a decent mapping. (I didn't use RFC 9651 booleans because they are user-hostile.)
I'm interested in how parameters would be represented in JSON
", "time": "2025-06-23T14:09:02Z"}, {"author": "Martin Thomson", "text": "Timid Robot: yes, that would be a very good thing to explore
", "time": "2025-06-23T14:09:24Z"}, {"author": "Farzaneh", "text": "I am interested in time of collection vs. time of use
", "time": "2025-06-23T14:09:39Z"}, {"author": "Felix Reda", "text": "I think I don't understand enough about the attachment syntax to take useful notes, feel free to jump in.
", "time": "2025-06-23T14:10:58Z"}, {"author": "Timid Zehta", "text": "(#13) +1 to focusing on robots.txt and HTTP content-usage header
", "time": "2025-06-23T14:11:23Z"}, {"author": "Farzaneh", "text": "I think it should be user centric in a way that when the AI developer wants to follow the signal, it doesn't get contradicting and complex signals.
", "time": "2025-06-23T14:16:26Z"}, {"author": "Farzaneh", "text": "time of collection vs time of use: I can't locate this mentioned in the Attachment doc, has it been mentioned?
", "time": "2025-06-23T14:18:27Z"}, {"author": "Suresh Krishnan", "text": "@Farz, I think section 7.1 of Vocab is fairly straightforward for handling multiple signals. Thoughts?
", "time": "2025-06-23T14:18:58Z"}, {"author": "Farzaneh", "text": "I dont see that it talks about \"timing\" does it?
", "time": "2025-06-23T14:19:55Z"}, {"author": "Martin Thomson", "text": "Preferences only apply to use, not collection.
", "time": "2025-06-23T14:19:57Z"}, {"author": "Felix Reda", "text": "https://github.com/ietf-wg-aipref/drafts/blob/main/draft-ietf-aipref-attach.md
", "time": "2025-06-23T14:20:08Z"}, {"author": "Martin Thomson", "text": "https://ietf-wg-aipref.github.io/drafts/draft-ietf-aipref-attach.html is the output
", "time": "2025-06-23T14:20:25Z"}, {"author": "Leonard Rosenthol", "text": "@martin, because we are now say that the updated REP covers \"colllection\" and ai-controls covers use. Is that your point? (if so, I'm good with that!)
", "time": "2025-06-23T14:21:20Z"}, {"author": "Suresh Krishnan", "text": "@Farz, my response was to your first comment regarding \"contradicting and complex signals\".
", "time": "2025-06-23T14:22:06Z"}, {"author": "Farzaneh", "text": "ah thank you
", "time": "2025-06-23T14:22:40Z"}, {"author": "Thom Vaughan", "text": "The hackathon sounds like a great idea
", "time": "2025-06-23T14:23:45Z"}, {"author": "Bryan Newbold", "text": "I'm planning to do a hackathon project in Madrid doing AI prefs with a social web usecase (and sounds like I should post about this on the mailing list)
", "time": "2025-06-23T14:23:59Z"}, {"author": "Bradley Silver", "text": "@Martin - the charter says - The AI Preferences Working Group will standardize building blocks that allow for the expression of preferences about how content is collected and processed for Artificial Intelligence (AI) model development, deployment, and use.
", "time": "2025-06-23T14:24:28Z"}, {"author": "Martin Thomson", "text": "For folks who are interested, I have a rust implementation of the grammar that needs updating.
", "time": "2025-06-23T14:24:33Z"}, {"author": "Bradley Silver", "text": "Therefore - Collection?
", "time": "2025-06-23T14:24:43Z"}, {"author": "Timid Zehta", "text": "thank you all <3
", "time": "2025-06-23T14:24:55Z"}, {"author": "Farzaneh", "text": "thanks all.
", "time": "2025-06-23T14:24:59Z"}, {"author": "Aditya Kanteti", "text": "Thanks everyone
", "time": "2025-06-23T14:25:04Z"}, {"author": "Thom Vaughan", "text": "https://datatracker.ietf.org/meeting/123/agenda also this is the preliminary agenda for IETF 123 Madrid
", "time": "2025-06-23T14:25:07Z"}, {"author": "Sonia Cooper", "text": "Thanks
", "time": "2025-06-23T14:25:10Z"}, {"author": "Martin Thomson", "text": "It could be AND in terms of set intersection. That is, it only applies to content that has been acquired and processed, both.
", "time": "2025-06-23T14:25:14Z"}, {"author": "Bryan Newbold", "text": "thanks all!
", "time": "2025-06-23T14:25:15Z"}, {"author": "Jessie Stricchiola", "text": "thanks all
", "time": "2025-06-23T14:25:16Z"}]