How to Extract Text From a Screenshot (Every OS)

You’ve got a screenshot with text you can’t select — an error message, a chat, a slide, a receipt, a table someone sent as a picture — and you need those words as real, editable text. The good news is your operating system can probably already do this, no download required. This guide gives you the fastest built-in method on every platform first, honestly, and then covers the cases where the built-in tools quietly stop being enough.

macOS: Live Text is built in

If you’re on a Mac running a recent macOS, Live Text reads text inside images system-wide.

• Quick Look: Select the screenshot file in Finder and press Space. In the preview, click and drag across the text — it selects like normal text. Copy it.
• Preview or Photos: Open the image, then click and drag over any text to select and copy it.
• Right in the screenshot thumbnail: After you take a screenshot, the thumbnail that appears bottom-right lets you select text before it even saves.

No app, no setting to enable in most cases. If dragging doesn’t select text, the language may not be supported or the image may be too low-resolution — more on that below.

Windows: Snipping Tool and PowerToys

Windows has two solid built-in-ish routes.

• Snipping Tool (Text Actions): On Windows 11, open Snipping Tool, capture or open your screenshot, and use Text Actions to extract and copy the text right there. It’s the no-install answer for most people.
• PowerToys Text Extractor: If you install Microsoft’s free PowerToys, it adds a shortcut (default Win + Shift + T) that lets you drag a box over anything on screen — a window, a video frame, a screenshot — and the text lands on your clipboard instantly. Once you’ve used it, it’s hard to go back; it’s the fastest screen-to-text on Windows.

Google Lens: works from any browser

Don’t want to install anything and not sure what OS conveniences you have? Google Lens reads text in an uploaded image and lets you copy it. Open Lens in a browser or the Google app, add your screenshot, and select the detected text. It handles a wide range of languages and is a reliable fallback across devices.

iPhone and iPad: Live Text again

iPhones have the same Live Text as the Mac. Open the screenshot in Photos, then press and hold on the text — a selection appears with grab handles, plus a Copy option. You can also tap the Live Text icon (bottom-right of the photo) to select everything at once. It works in the camera, in Photos, and in most places an image shows up.

Android: Google Lens and Circle to Search

On Android, Google Lens is built into Photos and the Google app — open the screenshot, tap the Lens icon, and select the text to copy. Many recent phones also have Circle to Search: long-press the home button or gesture bar and circle the text on screen. Both are no-install on most modern Android devices.

So when do the built-in tools fall short?

For a single screenshot with clean, printed text in a common language, the methods above are genuinely all you need — use them. But there are real situations where they get slow, awkward, or aren’t allowed at all:

• Batches. Extracting text from one screenshot by hand is fine. Doing it for thirty is miserable. The native tools are one-at-a-time, drag-and-select affairs with no way to queue a folder.
• Tables and structured layouts. Live Text and friends are built to grab a run of text, not to preserve a grid. Pulling a table out of a screenshot often gives you a jumbled stream of cells you then have to reassemble.
• Privacy at work. A screenshot of an internal dashboard, a customer record, a contract, or anything under an NDA is exactly the thing you should not upload to a web service. Google Lens and many “OCR online” sites send your image to a server. For sensitive captures that’s a non-starter.
• No install rights. On a locked-down work or school machine you may not be able to add PowerToys, and the OS OCR may be disabled or missing on an older version. You need something that runs with nothing to install.
• Older or minimal systems. Not every machine has the latest macOS or Windows 11 conveniences.

The fallback: OCR that runs in your browser

When the built-in path doesn’t fit, our image-to-text tool covers the gaps without asking you to trust a server. You drop a screenshot (or several) onto the page and the optical character recognition runs entirely inside your browser tab. A small recognition model downloads once and caches; after that, every image is read locally. The picture is never uploaded — you can confirm it in your browser’s network tab, where you’ll see the model load but never see your image leave.

That design is what makes it the right tool for the awkward cases:

• Batches: add several screenshots at once and they’re read in order, instead of one tedious drag-select at a time.
• Privacy at work: because nothing is transmitted, a screenshot of a private dashboard or a confidential document stays on your machine. There’s no server that ever holds a copy — the guarantee is structural, not a promise in a policy.
• No install rights: it’s a web page. Nothing to download, nothing to approve with IT, works on a locked-down machine.
• Editable output: the extracted text lands in an editable pane you can copy or download as a .txt, so you can fix a stray character before you save.

It reads printed and typeset text well — receipts, screenshots, book pages, forms, slides, labels. Handwriting is the one thing on-device OCR still struggles with, and it’s worth being honest about that rather than overselling it.

What about a screenshot of a PDF page?

If your “screenshot” is really a page from a document you have as a PDF, don’t screenshot it at all — extract the text from the PDF directly. That’s cleaner and preserves layout far better than any OCR pass over a picture. Our PDF to text tool handles that, and it also runs in your browser.

Quick reference

• One clean screenshot, common language: use your OS — Live Text (Mac/iPhone), Snipping Tool or PowerToys (Windows), Lens or Circle to Search (Android). Free, instant, already installed.
• A batch, a table, sensitive content, or no install rights: use in-browser OCR so nothing uploads and you can process several at once.
• It’s actually a PDF: pull the text from the PDF instead of screenshotting it.

The words in a screenshot are almost never truly stuck. Reach for the built-in tool first, and when it doesn’t fit the job, reach for OCR that keeps your capture on your own device.