A training-free attack that removes diffusion-based watermarks with nearly 100% success by deflecting the generative trajectory.
April 1, 2026
Original Paper
SHIFT: Stochastic Hidden-Trajectory Deflection for Removing Diffusion-based Watermark
arXiv · 2603.29742
The Takeaway
It challenges the conventional wisdom that generative watermarks are robust to removal, proving they can be stripped without loss of semantic quality or watermark-specific knowledge, rendering current paradigms vulnerable.
From the abstract
Diffusion-based watermarking methods embed verifiable marks by manipulating the initial noise or the reverse diffusion trajectory. However, these methods share a critical assumption: verification can succeed only if the diffusion trajectory can be faithfully reconstructed. This reliance on trajectory recovery constitutes a fundamental and exploitable vulnerability. We propose $\underline{\mathbf{S}}$tochastic $\underline{\mathbf{Hi}}$dden-Trajectory De$\underline{\mathbf{f}}$lec$\underline{\math