[webai] [PROPOSAL] Tracking and Exploring AI and Accessibility (#5)

ruoxiran has just created a new issue for https://github.com/w3c/webai:

== [PROPOSAL] Tracking and Exploring AI and Accessibility ==
## Summary 

This issue is created to follow up on the discussions from the [Web & AI IG meeting on February 9, 2026](https://www.w3.org/2026/02/09-webai-minutes.html#92d6). Based on [the presentation](https://www.w3.org/2025/04/AC/talk/atkinson#talk) by Matthew @matatk and the subsequent group discussion, several key areas require further exploration and tracking: 

- Use of Accessibility Metadata: There is a need to be cautious about how machines use rich accessibility attributes. We must ensure that this information is not misused or used to create a false sense of accessibility where gaps still exist.
- Generative AI Hallucinations: While generative AI has the potential to make the web more accessible, we must address the systemic impact of AI hallucinations on the reliability of these descriptions, concerns about the effectiveness of flagging content specifically as "AI-generated".  A flag might give a false impression of accuracy, or conversely, the absence of a flag might wrongly imply human origin. 
- National level of around standards relating to AI and Accessibility.

Context from the meeting, the discussion also touched upon:

- AI Assurance Principles: Incorporating AI assurance principles was proposed as a way to evolve our work and surface accessibility more effectively. 
- WebMCP accessibility considerations: https://github.com/webmachinelearning/webmcp/issues/65
- Accessible Provenance: How to ensure that provenance metadata is available and accessible to users across different contexts.

In addition to the discussions highlighted above, there are several other venues within W3C talked about AI and accessibility, including:

- [AI impact on the Web](https://github.com/w3c/ai-web-impact) and its [related issues](https://github.com/w3c/ai-web-impact/issues/24)
- RQTF's work on [Accessibility of machine learning and generative AI](https://w3c.github.io/ai-accessibility/)

---

## Areas of Interest  (Optional)  
Which aspects might be relevant? (Check any that apply)

- [ ] AI ethics, responsibility, and impact
- [ ] AI and Web platform technologies  
- [ ] AI Accessibility  
- [ ] Security / privacy / trust  
- [ ] Other: ____________________

---

## Discussion Questions  (Optional)  
Questions, thoughts, and further discussion are very welcome, and if needed, we can continue the discussion in a regular meeting.

Please view or discuss this issue at https://github.com/w3c/webai/issues/5 using your GitHub account


-- 
Sent via github-notify-ml as configured in https://github.com/w3c/github-notify-ml-config

Received on Tuesday, 10 February 2026 09:05:02 UTC