Web pages have long involved a tradeoff: standard HTML elements can be clicked, selected and read by screen readers but are hard to warp visually, while canvas can draw anything yet treats its content as a flat image. HTML-in-Canvas places real page elements inside canvas graphics while keeping them interactive. Google is developing it as a web standard that preserves both interactivity and accessibility.
Felix Rieseberg, who works on Claude at Anthropic, wrote on X that Claude can now build "bizarre & beautiful sites" with the feature and said it reminds him of the Flash era. A demo page shows 15 examples, including sites rendered as knitting patterns, 3D cubes and exploded layer views, and the text stays highlightable and copyable.
The feature works in Chrome and Edge starting with version 157 and in Claude's desktop apps from version 5.11492. Safari and Firefox do not support it yet.
Why it matters if you build agents
Agents that generate front-end pages can now produce unusual visual layouts without losing text selection, links or screen reader access. Developers testing generated sites should check browser support, since Safari and Firefox lack it for now.
Users have to request HTML-in-Canvas explicitly, so agent prompts and workflows that expect it need to say so.
Get stories like this every weekday
One short email with the day's agent tooling news.