When the “Confluence Page Viewer” replaced the “Confluence Page Gadget”, it introduced a change where the gadget now requires an active session on Confluence to load images and icons.

In Jira 9.12, the Confluence Page gadget was provided by Confluence itself and likely used a server-side proxy mechanism to fetch all page content, including images. This meant everything was retrieved through the application link on the server side, so your browser didn’t need its own session to Confluence.

In Jira 10.X the page text and HTML content are still fetched via the application link, which continues to work well. However, the images within that HTML now use direct URLs pointing to Confluence instance.  Because of which the browser tries to load those image URLs, it needs a valid Confluence session and if not it loads broken images.