Content injected to PDF rendering contexts could, in many places, include HTML content including tags. If the src attribute of these images pointed to an URL, the PDF rendering engine would download the image from that place and display it, thereby leaking information about the rendering server and possibly creating an SSRF vector in the local network.