The scanner is creating pages that are not ADA compliant

I ran the scanner on a new project today. It fixed one of my html files and also added three html error pages. I merged those PRs. (It also tried to add a test script, js and package.json it appears to have used to validate it’s fixes, but I did not merge those.) I reran the scanner to verify the fixes, and it found compliance issues with the three error pages copilot created in the previous scan.

For example the left side of this screenshot is the original html the scanner created, the right is the issue found when I ran the scanner again:

Start by reproducing the scanner run that creates the three HTML error pages, then inspect those generated pages alongside the accessibility issue shown in the screenshot. The unmerged test script, JavaScript file, and package.json may indicate how the previous fixes were validated. Done means the generated error pages pass a subsequent scanner run without reporting the same compliance issues.

索引モデルが issue の本文から書いたものです。