Anthropic engineer Felix Rieseberg said he enabled HTML-in-Canvas in Claude's desktop apps, allowing generated pages to place selectable text within canvas graphics and 3D effects.
HTML-in-Canvas keeps text as ordinary page elements while drawing their appearance into a canvas. Google says browser functions such as Find-in-Page, translation and autofill can work on that text; graphics drawn as pixels, such as chart slices, do not gain those functions.