Which Copy Leaked? Per-Recipient Watermarks for Unreleased Webtoon Pages
To find out which copy of a webtoon episode leaked, each recipient has to receive a different copy. If the same file goes to every partner, a leaked image can be matched to the work but not to the person or company it was sent to. A per-recipient watermark addresses this by placing a different invisible mark in each copy before it leaves, so the mark found in a leaked image points back to one copy.
Don't webtoon platforms already trace leakers?
They do, mostly at the reader level. On October 25, Korean daily Edaily reported that Kakao Entertainment's anti-piracy team, P.CoK, links invisible watermarks to its in-house system, DarkChaser, to identify the user accounts that took content out. The team said that once a leaker is confirmed, it can go from leak to account sanction in as little as an hour. IT Chosun describes Naver Webtoon's Toon Radar in similar terms: it puts identification information that cannot be seen into webtoon images to trace illegal leakers.
As reported, both systems work on identifying the reader account behind a leak of a published episode.
Why isn't blocking screenshots enough?
Because copy protection can be hard to rely on alone. IT Chosun found on October 13 that free episodes on Naver Webtoon and webtoons on KakaoPage could be captured on a PC with a single Print Screen key, and that the scope of capture restrictions differed by platform. In a parliamentary audit demonstration reported on October 12, the latest free episode was saved as images and uploaded to another site in under a minute. The head of the Korea Copyright Protection Agency answered that differences between apps and web browsers mean copy protection works in some places and not in others.
P.CoK's Kwon Young-guk put it plainly to Edaily: it is hard to stop content leaks 100% online.
Which part of the pipeline does the rights holder own?
The part before publication. This is our reading, not something the articles state: a studio or IP holder may send unpublished episodes to platforms, overseas publishers, and translation or lettering partners. Those copies exist before any reader account does, so a platform's reader-level mark may not cover them.
Where the leak happened | Mark that needs to be in the file | Question it can answer | Who checks |
|---|---|---|---|
Unpublished pages sent to platforms or translators | A different watermark for each recipient | Which recipient's copy is this? | Rights holder |
Published episode viewed by readers | The platform's per-user mark | Which account did it come from? | Platform |
Re-uploads that have already spread | The original copy's mark, if it survived | Which copy was the first to leave? | Rights holder and platform |
In an earlier guide we compared watermarks, C2PA and fingerprints. A fingerprint matches the work, and every recipient's copy has the same fingerprint, so a mark that differs per copy is what lets copies be told apart.
What does tracing a leaked copy look like in practice?
Here is a hypothetical scenario. Studio A sends the next episode of a new series to a domestic platform, an overseas publisher and a freelance translator before release. Before sending, it embeds an invisible watermark with a different value in each copy and keeps a record of which value went to whom.
The day before release, the episode shows up on a piracy site. Studio A runs watermark detection on the posted images, and the value matches the translator's copy. Alongside the takedown request, the studio now knows which working environment to look into. If the mark had pointed to the platform's published version instead, the next step would be to ask that platform to check its accounts.
Tools cover parts of this flow. At MuseBlossom, Image Defence embeds invisible watermarks in images and issues and validates C2PA manifests. Using a different value per recipient to confirm which copy leaked, and packaging the result as evidence for a takedown notice, is offered as a DMCA and leak-tracing proof of concept (PoC). Finding where the content has been posted can be tested with K-TRACE, the leak monitoring feature of Contents Defence, as a PoC, including for webtoons.
What can't a per-recipient watermark tell you?
It tells you which copy, and stops there. Who inside the receiving company sent it out, and whether it was deliberate, has to be worked out from contracts and handling records. A copy can also leak without the recipient knowing.
The mark also has to survive in the posted image. If the images were heavily edited or redrawn before upload, detection results may differ.
FAQ
If the platform already watermarks episodes, does the rights holder need its own mark?
A platform's mark points to that platform's reader accounts. Pages that go to several partners before release need a mark based on who the rights holder sent them to, or the copies cannot be told apart.
Will an invisible watermark change how the artwork looks?
An invisible watermark is designed not to be noticed. Art styles, colors and textures vary, so it is worth testing on a few real pages first.
Once the copy is identified, is the case settled?
No. Identifying the copy is the starting point. Responsibility depends on contract terms and on the records of how the file was delivered and handled.
Want to test invisible watermarks on a few of your pages first? Ask us about Image Defence. go.museblossom.com/image