What it does
| View | What it shows |
|---|---|
| Reader | The document rebuilt as HTML — from the author's own structure tags where the PDF has them, and from inferred structure where it does not. |
| Original | The original PDF pages, which work even when no text can be extracted. |
| Markdown | Copyable text: either what the Reader shows, or the parser's raw output. |
Normally you open a PDF and press the toolbar button. Switching
Opening PDFs to the Reader in the settings makes addresses
ending in .pdf open here directly instead; the browser asks for
access at the moment you switch it on, and gets it back when you switch it off.
The principle
It does not assert what cannot be confirmed by looking. An image with no description is reported as having none. Machine-read text is labelled as machine-read every time it is read. When a page yields no text, the extension finds out whether it is an image, text drawn as outlines, or genuinely empty — and says which.
| Source | How it is shown |
|---|---|
| Written by the author | Read out as the image's alternative text; shown on screen when the image cannot be |
| Generated by the authoring tool | Labelled as automatically generated |
| Generated by this extension | Labelled as an AI-generated description |
| Missing | Reported as missing |
Features
| Feature | What it does |
|---|---|
| Heading navigation | Jump between headings with H and 1–6, the same as any web page. |
| List and table navigation | Real HTML lists, and tables with row and column headers. |
| Link navigation | Every link stays a link, reachable from the browser's own links list. |
| Image descriptions (Chrome only) | On-device AI description for images with no alternative text, always labelled as AI-generated. |
| OCR for scanned pages (Chrome only) | On-device text recognition for pages with no extractable text, labelled as machine-read wherever it is read. |
| Text and layout | Text size, line spacing, column width and typeface, including a bundled dyslexia-friendly face. |
| Theme | Light, dark, or follows your system setting. |
| Interface language | English or Japanese — independent of the document's own language. |
| Two readings of structure | The author's own tags where present, or inferred structure — switchable, so you can check one against the other. |
Privacy
PDFs are parsed inside the browser. In ordinary use no document is sent to any server. The optional image descriptions and OCR use the browser's own on-device model, which runs on your machine.
It asks for five permissions at install in Chrome, four in Firefox, and none of them is a host permission: installing it grants access to no site. One optional host permission exists, asked for only when you do something that needs it: one site, when a PDF on it could not otherwise be read, or every site, when you switch Opening PDFs to the Reader — handed back when you switch it away.
What's new
Version 0.2.4 (): small figures the author described are shown as pictures, and a figure whose alternative text is empty is no longer treated as one the author marked decorative.
What's new: every version so far
Getting it
Accessible PDF View on the Chrome Web Store
It needs Chrome 147 or later, or Edge 147 or later.
With Chrome's Enhanced Safe Browsing turned on, installing shows “This extension is not trusted by Enhanced Safe Browsing.” The status is earned by a developer's record of publishing within the Chrome Web Store's policies, and Google says that for new developers it generally takes a few months. Accessible PDF View was first published in September 2026, so until then the warning appears.
To install it anyway, choose “Continue to install”. The permissions it asks for and what it does with your data are set out under Privacy, and the source code is on GitHub. Google explains the warning in the Chrome Web Store Help article “Install and manage extensions”.
Firefox is not listed yet. A listing on its add-on site is still to come, and until then the Firefox build — Manifest V3, like the Chrome one — can be loaded from source in Firefox 144 or later. Image descriptions and OCR use Chrome's built-in AI and are available there only.