An accessible, or tagged, PDF carries a hidden structural layer, headings, paragraphs, lists, tables, and reading order, separate from its visual layout, along with alt text on images and correctly labeled form fields. Without these tags, a PDF that looks perfectly normal on screen can be entirely unreadable to a screen reader, which sees only an unstructured wall of text or, in scanned documents, no text at all.
PDF accessibility is a distinct discipline from web accessibility, with its own set of tools and standards (PDF/UA), and it is one of the most common gaps organizations discover once they start auditing beyond their website, since PDFs published years ago were often never checked at all.