Semantic watermarks in LDMs have an irreducible geometric distortion floor from proxy-target model mismatches that limits forgery fidelity and supports scheme-agnostic detection via global drift and local deformation.
arXiv preprint arXiv:2506.06018 , year=
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CR 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Rethinking Forgery Attacks on Semantic Watermarks in Black-Box Settings: A Geometric Distortion Perspective
Semantic watermarks in LDMs have an irreducible geometric distortion floor from proxy-target model mismatches that limits forgery fidelity and supports scheme-agnostic detection via global drift and local deformation.