W3C

– DRAFT –
COGA Research Subgroup Meeting

23 June 2026

Attendees

Present
Becca_Monteleone, julierawe, LenB, Rashmi
Regrets
-
Chair
LenB
Scribe
julieq, julierawe

Meeting minutes

<LenB> present_

Link to COGA Trigger Warnings module: https://docs.google.com/document/d/19RdKrd2aUJkZcm9FmIOvZ2jfn9JgYpFcqT5mubyQUAQ/edit?tab=t.0#heading=h.nidxqwliwmpj

Link to Len's Trigger warnings top-up research: https://docs.google.com/document/d/1XWmikm5Ein5CQ02knEJ5iGzO1YJMilTMBJ_aLVCTzs0/edit?tab=t.rfkd7gpxr2qm#heading=h.h8q514yzmd7o

LenB looked at our mental health articles database and reviewed the instructions

LenB recapped that he searched Google Scholar for "online triggers" since 2022

LenB and @Rashmi signed up to look at different papers

LenB decided would take a long time to look at one paper at a time.

LenB loaded COGA's trigger warning paper into Claude and then pasted the first 3 studies into Claude and asked it to figure out what is new /missing and are there tensions between those new papers and our module.

LenB There are 27 new papers for us to consider.

LenB switched from Claude to Gemini, which is better trained to work with Google Scholar.

LenB created an entry "New themes (not yet in our paper)"

LenB: What do you all think of my process?

LenB: I also read the 27 papers. It probably took me 45 minutes to an hour every other day for a few weeks to read all of them.

julierawe: It's really important to check for accuracy of the AI so it's great that you've read the studies.

LenB: Yes, also important to make sure it's keeping to our scope.

LenB: I used AI to help queue up important things to read. It helped me feel more confident that I was reading what I should read,

Becca_Monteleone: The only thing I would flag is doing multiple searches with multiple terms.

Becca_Monteleone: I was searching for "wayfinding" and "cognitive disabilities." This language is changing pretty rapidly and also looks different in different parts of the world.

Becca_Monteleone: Some of the modules have the search terms listed at the end.

LenB: The Trigger Warnings draft did not yet have a section like this.

Rashmi corrected this and showed the section in the draft that lists the search terms used so for for the draft.

Correction: The mental health research articles doc has a section that shows how to do this. This section still needs to be drafted for the trigger warnings draft.

Link to the "Search" instructions in the COGA doc titled "Mental health research articles": https://docs.google.com/document/d/1XWmikm5Ein5CQ02knEJ5iGzO1YJMilTMBJ_aLVCTzs0/edit?tab=t.0#heading=h.vipcku7wnabt

Becca_Monteleone: For this paper, I would search for "online triggers mental health"

Becca_Monteleone: I would search for:

--online triggers + mental health

--online triggers + mental illness

--content warning + online triggers

rash

LenB listed several terms in the "Top up for Triggers" tab of the doc, including "coping mechanisms" and "misuse"

Link to "Top up for Triggers" tab of the doc: https://docs.google.com/document/d/1XWmikm5Ein5CQ02knEJ5iGzO1YJMilTMBJ_aLVCTzs0/edit?tab=t.rfkd7gpxr2qm

LenB: The research identified a lack of a standard for how the triggers appear, how they're worded.

LenB: Identifying a lack of standards is important because it can held lead to identifying misuse.

ACTION: item: LenB and @ Rashmi to rerun the search with the new terms.

Rashmi: The process is very good, very effective. It helps us dive deeper.

Rashmi: My concern is whether human interpretation and AI interpretation are aligning.

LenB: That's a valuable question that is often hidden in the results. If a human has done the work and we trust ourselves to do the work.

LenB: You can sometimes see phrases that AI is using rather than a human. I sometimes edited those phrases so humans wouldn't dismiss them.

LenB: Sometimes I made the AI more concise.

LenB: Example of platform design. That's a section I shortened. https://docs.google.com/document/d/1XWmikm5Ein5CQ02knEJ5iGzO1YJMilTMBJ_aLVCTzs0/edit?tab=t.rfkd7gpxr2qm

LenB: I did leave some things in, I did edit some things. I didn't want to put in a lot of comments because it would make it hard to read the notes.

LenB: We need enough information to make decisions and move on to the next phase.

julierawe: It sounds like you took a close look at the AI notes and scrutinized them based on you having read the papers. Is that a good summary?

LenB: Yes.

LenB: I used AI but I also read the papers without AI.

LenB: I should write up this process in this document.

ACTION: item: LenB to add to the doc an outline of what he did that has more detail.

LenB: I could use this process for all of these new search terms we came up with today.

LenB: Our paper will be stronger because of it.

Rashmi: You have done a massive job. This process is very effective. Tremendous work, so hats off to you!

Rashmi: You made it simple.

Research top-off efforts for the Trigger Warnings issue paper

New themes (not yet in our paper

Link to "New themes" section of the top-up: https://docs.google.com/document/d/1XWmikm5Ein5CQ02knEJ5iGzO1YJMilTMBJ_aLVCTzs0/edit?tab=t.rfkd7gpxr2qm#heading=h.eureqphebh15

julierawe noted several new themes seem focused on social media. Are news sites also included?

LenB added a note about this. These are the results that showed up.

julierawe: Sounds like we could deepen the module's discussion of social media, but would be good if we could make sure we aren't missing research about other kinds of content that isn't social media.

LenB noted the "Tensions" section pointed to a new paper that said people with trauma histories are no more likely to avoid warned content than other people.

LenB: I don't think we can land on one standard and shared Lisa's story about her dad and smoking.

Link to "Tensions" section: https://docs.google.com/document/d/1XWmikm5Ein5CQ02knEJ5iGzO1YJMilTMBJ_aLVCTzs0/edit?tab=t.rfkd7gpxr2qm#heading=h.5i0517pnnjje

Tensions between new research and our paper

Link to "Research previously / also considered": https://docs.google.com/document/d/1XWmikm5Ein5CQ02knEJ5iGzO1YJMilTMBJ_aLVCTzs0/edit?tab=t.rfkd7gpxr2qm#heading=h.f8ycvag0uqbu

LenB: Highlighted one paper in this section that notes the importance of cultural framing.

LenB: We may want to run a search on cultural framing and that might become a new theme.

In the "Tensions" section, there is a section that says we may be overstating the drawbacks.

LenB: A new study by Bridgeland contradicts a study we used for our research module.

LenB: The 2024 meta-analysis supersedes those with a larger evidence base and a more nuanced conclusion - warnings are described as "trivially helpful" rather than actively harmful.

Rashmi: I'm hearing from our discussion that customization is the most helpful.

Rashmi: There is one pattern we drafted for COGA on the presentation of trigger warnings and triggering content:

https://docs.google.com/document/d/1Itskr4srybRQa7hyIqMRiOPyjFyKB9apHYWdvGRb6gE/edit?tab=t.0#heading=h.ko4kl8hexve2

Rashmi: We want to give users the choice.

Rashmi: I want to learn more about the tensions so we can concentrate more on that.

LenB: It's looks like the next part of our agenda is premature. We need a week or so to think about whether this top-up research affects our GitHub issues.

Research previously / also considered

LenB: It's looks like the next part of our agenda is premature. We need a week or so to think about whether this top-up research affects our GitHub issues.

ACTION: item: LenB and Rashmi will meet to figure out how we will divide up the additional papers that surface when we search for these new terms.

Rashmi: What is our urgency?

julierawe: WCAG 3 is a moving target. If you need an extra week to think about GitHub issues, you can certainly have some extra time. But the sooner we can get the issue filed, the better.

How does the new research affect our WCAG 3 review GitHub issues for Animation and Movement?

LenB: The big questions are is this a good process that we want to adopt? How do we evaluate papers to say something is worthy, not worthy, or out of scope? With this process, we can look at the subject as a group and decide if this makes it into our module or not. That's a big team question, not a one-paper question.

LenB: I also want to explore if this meeting time still works for Abi and Eric. Might later in the day be better? Maybe we wait until after the summer to evaluate this scheduling question.

Group discussed what a human needs to do check the AI.

LenB: You can check one AI assessment by checking it with another AI tool.

LenB: It also depends on the humans' style and preference for reading documents. We're always going to have variance in the way a researcher does their work.

LenB: We could add a step for using one AI tool to check another AI tool.

LenB: The methods a human uses to review a document—could be reading the summary/table of contents. Could be reading a section, could be reading the whole paper. We don't have these methods listed as a team. The new tool has forced us to think about these things.

Rashmi: Is the AI tool including the papers we have cited?

LenB: We could have the tool strip out those citations so we aren't looking at already visited territory.

Summary of action items

  1. item: LenB and @ Rashmi to rerun the search with the new terms.
  2. item: LenB to add to the doc an outline of what he did that has more detail.
  3. item: LenB and Rashmi will meet to figure out how we will divide up the additional papers that surface when we search for these new terms.
Minutes manually created (not a transcript), formatted by scribe.perl version 248 (Mon Oct 27 20:04:16 2025 UTC).

Diagnostics

Maybe present: Correction

All speakers: Becca_Monteleone, Correction, julierawe, LenB, Rashmi

Active on IRC: Becca_Monteleone, julierawe, LenB, Rashmi