How to Copy Text That Cannot Be Copied

“Cannot be copied” covers four very different situations, and the fix for one is useless for the others. Work out which one you are in first — it takes about five seconds and saves you from installing something you do not need.

Where the text is	Why selection fails	The free fix
A web page	CSS user-select: none	Reader mode, print preview, or view-source
A PDF that highlights but won’t copy	Permissions flag in the file	Print to a new PDF, or re-open in a different viewer
A PDF where nothing highlights	It is a scan, with no text layer	OCR — there is genuinely nothing to select
A native app (banking, Telegram, a game)	The app never built a selection handler, or set a secure flag	Screenshot, then text recognition

Web pages: the block is cosmetic

If a site stops you dragging across a paragraph, it is almost always one CSS declaration on the page body: user-select: none.

That property tells your browser how to respond to a drag. It does nothing to the text itself, which is sitting in the page source exactly as it always was. Any route that re-renders the content escapes it:

• Reader mode. Firefox and Safari both extract the article body into their own view, and your CSS rules do not travel with it. This is the cleanest option and it keeps paragraph breaks intact.
• Print preview. Press Ctrl+P (Cmd+P on a Mac) and choose “Save as PDF”. The print stylesheet almost never carries the restriction.
• The console. Open developer tools and run document.designMode = 'on'. The page becomes an editable document and everything selects normally.
• View source. Ctrl+U shows you the raw HTML, which was never subject to the rule.

Right-click menus are blocked with JavaScript rather than CSS, and the same routes defeat those too.

PDFs: two different problems wearing the same coat

Try to highlight a line. What happens next tells you which problem you have.

If the text highlights but Copy produces nothing, the file carries a permissions flag. This is part of the PDF specification, and it works on the honour system: the flag asks viewers not to allow extraction, and mainstream viewers comply voluntarily. Printing the document to a new PDF usually produces a file without the flag, and that copy behaves normally.

If nothing highlights at all — your cursor just draws a rectangle — the page is an image. Someone scanned it, or exported it from a photo. There is no text layer to select, so no browser trick will help, and OCR is the only route: screenshot the page and run it through our tool for copy-blocked text. That is not a restriction; it is an absence.

Native apps

Inside an app, selection exists only if a developer built it. Banking apps routinely leave account numbers unselectable, not to stop you but because nobody wired up the gesture.

A smaller set of apps goes further and marks their windows as secure, which on Android means the system screenshot returns a black image. Telegram does this in channels with content protection enabled; the details and the workarounds are in copying text from a Telegram private channel.

For everything else, screenshot and read. On an iPhone, open the screenshot in Photos and tap the text-selection icon; Live Text handles it. On Windows 11, the Snipping Tool has a Text Actions button that recognises the capture locally and copies the result. On a Mac, Preview and Quick Look select text in images directly. These are free, already installed, and good enough most of the time — the full walkthrough is in copying text from a screenshot.

When the built-in tools are not good enough

Three cases where the free route reliably disappoints:

Long unbroken strings. An IBAN, a licence key, a wallet address, a 16-digit card number. Recognisers lean on language models to resolve ambiguous shapes, and a random string gives them nothing to lean on. One wrong character and the whole thing fails, silently.

Low-contrast interface text. Game HUDs, dark-mode terminals, semi-transparent overlays. Recognition quality falls off a cliff when the glyphs are small and the contrast is thin.

Anything you cannot proofread. Live Text drops text straight onto the clipboard. You paste it, it looks plausible, and you find out three steps later that a 0 became an O.

That last point is the argument for a tool that shows its work. Our copy-protected text tool puts the recognised characters into an editable box, so you can compare them against the screenshot before anything reaches your clipboard. For anything with a check digit or a fixed length, that one look is worth more than any accuracy claim. If you are dealing with digits specifically, why OCR gets numbers wrong explains which confusions to look for and why, and getting numbers out of an image into a spreadsheet covers the part where Excel quietly mangles what you pasted.

The short version

• Identify the block first: CSS, a PDF permissions flag, a scanned image, or a native app.
• On the web, user-select: none is cosmetic — Reader mode, print preview, view-source and document.designMode all walk straight past it.
• A PDF that highlights but won’t copy has a permissions flag; a PDF that won’t highlight at all is an image and needs OCR.
• In apps, screenshot and use the recognition already on your device: Live Text, Snipping Tool Text Actions, or Preview.
• Use a tool with an editable output box whenever the string is long, random, or something you cannot afford to get wrong.