Skip to content
SafeOCR.com

← All articles

How to Copy Text You Can't Select

SafeOCR TeamPublished June 27, 2026Updated July 5, 2026
A line of text being lifted out of an image as selectable text next to a copy icon.A line of text being lifted out of an image as selectable text next to a copy icon.

Text that won’t select falls into three common buckets: it’s baked into an image (a screenshot or photo), it’s a scanned PDF page with no text layer, or it’s on a page that has selection deliberately switched off. All three need the same tool, not a different browser trick — an OCR engine that reads the pixels and hands back real, copyable text.

Every browser lets you drag-select and copy text — until the text is not really text at all. If clicking and dragging over words selects nothing, or selects the entire page as one block, you have run into one of a few common situations, and all of them have the same fix: recognize the characters first, then copy the result instead of the pixels.

Case 1: the text is inside an image

Screenshots, photographed documents, and scanned pages are pictures. There is no text layer underneath for your cursor to grab, no matter how carefully you drag. This is the most common version of the problem, and it is what Copy Text from Image is built for: drop the image in, and every line the engine detects becomes a clickable box drawn over the picture. Click the box over the line you want and only that text is selected and copied — no careful dragging over pixels required. If the image started life as a screenshot specifically, the fastest route in is pasting it straight from your clipboard rather than saving it as a file first — see extracting text from a screenshot for that shortcut.

Case 2: a “flattened” or scanned PDF

Some PDFs look like documents but are really a stack of page images wearing a PDF file extension — common with faxed or scanned paperwork, and sometimes true of only some pages in an otherwise normal document. Selection tools in your PDF reader have nothing to grab onto because there is no text underneath, only a picture of it. Running each page through OCR first turns that picture back into real, copyable text; a full walkthrough of that process, including what to expect from multi-page files, is in the scanned PDF guide.

Case 3: a site or app with selection disabled

Some pages disable text selection with CSS, or wrap content in a canvas element instead of real text nodes — a deliberate choice, often aimed at discouraging casual copying. A screenshot of the on-screen content sidesteps that restriction: capture what is visible, then recognize the text from the image the same way you would any screenshot. (Whether copying that content is appropriate is a separate question from whether it is technically possible — use good judgment about what you have the right to reuse, particularly for anything under copyright or behind a paywall.)

Case 4: text baked into a graphic on purpose

Memes, exported chat bubbles, quote cards, and infographics often render their text as part of the image file rather than as real characters, specifically so it looks consistent everywhere it gets shared. There was never a text layer to select in the first place — the same recognize-then-copy approach applies here too, and it works regardless of why the text ended up as pixels rather than characters.

Why box-select beats a raw text dump

A busy image — a receipt, a business card, a cluttered screenshot — usually has more in it than the one line you actually want. Rather than handing you one undifferentiated block of text to hunt through, each detected line stays individually clickable, so pulling out a single confirmation code, phone number, or address is a single click rather than a search-and-select chore. Need everything? Copy all still grabs the full text in reading order, which matters when what you actually want is the whole thing pasted into a note or a spreadsheet rather than one value.

And it stays private

Because recognition runs on-device with WebAssembly and WebGPU, none of this requires uploading the image anywhere — which matters when the “text you can’t select” is something like an ID, a medical form, or an internal dashboard. The same on-device approach is what makes this practical for anything sensitive in the first place; see is online OCR safe for the full reasoning behind that.

We use Google Analytics and Google AdSense, which rely on cookies, only after you agree. Read our Privacy Policy