Skip to content
NewA calmer, faster ZeroHour

TetherMark ties image watermarks to one registered instance

Original title: TetherMark: Instance-Level Watermark Authentication Against Cross-Instance Transplantation

arXiv cs.CRby Jin Wei ()1 min read
AI summary · x-ai/grok-4.7Summaries by Vani AI

On 240 subjects it accepts 92.08% of legitimate cases, with attack success of 9.17% and 16.25%.

The paper describes cross-instance transplantation: an attacker moves valid watermark evidence from one registered image onto another image of the same category so the payload still decodes. TetherMark adds a relation check—secret patch pairs, a registered response, correspondence transport, and enrollment normalization—and accepts a mark only if both the message and relation checks pass. Across 240 test subjects it reports 92.08% legitimate true acceptance and the highest instance-binding balanced accuracy against WmForger and WMCopier.

Key facts
  • Attack success is 9.17% under WmForger and 16.25% under WMCopier.
  • The message check verifies an edit-robust payload.
  • The relation check tests consistency with the enrolled instance.
  • Standard decode tests do not show that evidence still belongs to the owner.
Explain this AI news with Vani AIWhat was announced, what is new and why it matters, in a new chat

From the original article

1 min read at arxiv.org

Robust watermarking is commonly verified by asking whether a hidden payload remains decodable after compression, geometric distortion, or generative editing. Localized and semantic extensions may additionally check whether an expected region or category is still present.

arXiv cs.CR

By Jin Wei · arxiv.org

Read the full article at arXiv cs.CR

A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.