What is a digital watermark?
Definition
A digital watermark is information embedded in an image, video, audio file or text, such as who made it or who owns it, so the content can be traced or checked later. It can be visible or hidden.
- Category
- Security
- Also called
- digital watermarking, watermark in a file
- Last updated
Reviewed by Anthon Wansland
Head of Product & Co-Founder
Serus is a privacy company with paid tools to find and remove your personal data. This page is free and based on official sources.
Sources checked: nvlpubs.nist.gov, merriam-webster.com
Last verified against the official page:
The word comes from paper. Merriam-Webster describes a watermark as a marking in paper, made by differences in thickness, that you can see when you hold the sheet up to the light. A digital watermark does the same job for files: it carries a small piece of information inside the content itself.
NIST's report on synthetic content (NIST AI 100-4) describes digital watermarking as embedding information into content such as an image, text, audio or video. That information can point to where the content came from or who owns it. NIST also separates detection methods into two kinds: blind ones, which can read the mark without the original file, and non-blind ones, which need the original to compare against.
Visible and hidden marks
A visible watermark is the faint logo or text across a stock photo. A hidden one is woven into the pixels, sound or wording so you can't see or hear it, and only a matching detector can read it. A hidden mark is only useful if someone has the detector for it.
Why it matters to you
Watermarks are now one way AI companies label images, audio and video their tools generate, so that others can check whether something is synthetic. They help, but they are not proof either way. A file with no watermark isn't necessarily real, and a watermark can be damaged when a file is edited or recompressed.
Watermark vs. related terms
- A digital fingerprint identifies a file or device from its own traits. A watermark is information added to the content on purpose.
- Metadata (such as camera details stored beside a photo) sits next to the content and is easy to strip. A watermark is built into the content.
- A deepfake is faked media. A watermark is one tool for flagging it.
Example
A photographer posts a portrait with a faint name across one corner. A stranger crops that corner off and uses the photo in an ad. The photographer's other copies also carry a hidden mark, so a detector can still match the cropped version to the original upload and show it came from them.
How to protect yourself
- 1
Treat a missing watermark as unknown
No mark doesn't mean a file is real, and a mark doesn't prove who made it. Check where an image or video came from before you trust or share it.
- 2
Watermark photos you share publicly
A visible mark across the image makes casual reuse harder. Keep an unmarked original offline so you can show it's yours.
- 3
Be careful about editing and recompressing
Cropping, filtering and re-saving can damage hidden marks. If you need the mark to survive, share the original file.
- 4
Check the source of surprising media
If a clip or image of you or someone you know looks off, find the original post and who published it first rather than relying on a mark alone.
See where your personal data is exposed
Serus finds it on data brokers and people-search sites and removes it.
Frequently asked
Can you see a digital watermark?
Sometimes. A visible watermark is a logo or text on top of the content. A hidden one is embedded in the file and needs a detector to read.
Can a digital watermark be removed?
Edits such as cropping or recompressing a file can weaken or destroy a watermark, which is why a watermark is a clue about a file's origin and not proof.
Is digital watermarking the same as metadata?
No. Metadata is stored beside the content and is easy to strip. A watermark is embedded in the content itself.
Sources
Something wrong or out of date? Tell us and we'll re-check it.