{"id":1994,"date":"2026-04-14T10:34:46","date_gmt":"2026-04-14T10:34:46","guid":{"rendered":"https:\/\/html5accessibility.com\/stuff\/?p=1994"},"modified":"2026-04-14T11:05:36","modified_gmt":"2026-04-14T11:05:36","slug":"last-week-in-wai-4","status":"publish","type":"post","link":"https:\/\/html5accessibility.com\/stuff\/2026\/04\/14\/last-week-in-wai-4\/","title":{"rendered":"LAST WEEK IN WAI #4"},"content":{"rendered":"
A stylish but illegal monkey wandering around in Toronto IKEA<\/figcaption>\"A<\/figure>\n

HTML in Canvas<\/a><\/h2>\n

I am interested in this as I always expected it to happen at some point, but appreciate that in this proposal (and experimental implementation) content drawn to the canvas is expressed in the canvas sub-DOM as HTML<\/a> which makes it usable by assistive tech. I don’t envisage, or found in limited testing, any issues for Screen Reader software. I don’t know how robustly it works with screen magnifiers, so I contacted a former colleague at Vispero and asked them to get their technical folk to review and feedback.<\/p>\n

Need a clarification of color requirements for images *with* text<\/a><\/h2>\n

The argument (by authority), put forward by one commenter on this discussion, I find bizarre and totally disagree with:<\/p>\n

The commenter makes the following claim:<\/p>\n

Image of text only applies to an image that only contains text.
\nOtherwise it is “text in an image” and follows other rules – like alt text.<\/p><\/blockquote>\n

source<\/strong>: https:\/\/github.com\/w3c\/wcag\/discussions\/5012<\/a><\/cite><\/p>\n

Take this as an example: An image containing only<\/strong> text fails WCAG 2.2 SC 1.4.3<\/a> as the text contrast in the image does not meet the contrast requirements
\n\"\"<\/p>\n

<img src=\"textimage.jpg\" \r\nalt=\"A stylish but illegal monkey wandering around in Toronto IKEA\"><\/pre>\n
If a separate image<\/strong> is added that is the subject of the text, the text still fails.
\n\"\"\"A<\/p>\n
<img src=\"textimage.jpg\" \r\nalt=\"A stylish but illegal monkey wandering around in Toronto IKEA\">\r\n<img src=\"stylishMonkey.jpg\" \r\nalt=\"A monkey wearing a puffy wool jacket\">\r\n<\/pre>\n
Now if the 2 images are combined (a single image with both text and picture) according to the commenter the image of text is no longer required to conform to WCAG 2.2 SC 1.4.3<\/a> contrast:<\/p>\n
\"\"<\/p>\n
<img src=\"textimage+stylishMonkey.jpg\" \r\nalt=\"A stylish but illegal monkey wandering around in Toronto IKEA\r\ntext above a picture of the stylish but illegal monkey wearing a \r\nwool jacket.\">\r\n<\/pre>\n
My reaction to this claim continues to be: what a load of bollocks<\/em><\/strong> which is why I have not commented on the discussion, as yet I find it\u00a0 impossible to fathom the reasoning, others may have a more measured response…<\/p>\n
Make of this what you will:<\/h2>\n
WCAG 3.0 Working Draft \u2014 Comment on Autonomous Agents as Primary and Proxy Users<\/a><\/p>\n
I have developed the AGAG \u2014 Agent Accessibility Guidelines \u2014 a dual-legibility framework extending WCAG 3.0 to explicitly include autonomous agents as first-class users. AGAG v1.0 organizes guidance around six pillars that map directly to the POUR principles: Detectable, Observable, Navigable, Economic, Verifiable, and Consent & Control.<\/p><\/blockquote>\n