VLA-Mark: A cross modal watermark for large vision-language alignment model

TL;DR

VLA-Mark is a novel cross-modal watermarking framework for vision-language models that embeds detectable watermarks without compromising semantic alignment, achieving high detection accuracy and robustness against attacks.

Contribution

It introduces a cross-modal watermarking method that preserves semantic fidelity and visual-textual coherence without requiring model retraining.

Findings

7.4% lower perplexity (PPL) compared to conventional methods

26.6% higher BLEU score indicating better language quality

98.8% AUC for watermark detection

Abstract

Vision-language models demand watermarking solutions that protect intellectual property without compromising multimodal coherence. Existing text watermarking methods disrupt visual-textual alignment through biased token selection and static strategies, leaving semantic-critical concepts vulnerable. We propose VLA-Mark, a vision-aligned framework that embeds detectable watermarks while preserving semantic fidelity through cross-modal coordination. Our approach integrates multiscale visual-textual alignment metrics, combining localized patch affinity, global semantic coherence, and contextual attention patterns, to guide watermark injection without model retraining. An entropy-sensitive mechanism dynamically balances watermark strength and semantic preservation, prioritizing visual grounding during low-uncertainty generation phases. Experiments show 7.4% lower PPL and 26.6% higher BLEU than conventional methods, with near-perfect detection (98.8% AUC). The framework demonstrates 96.1\% attack resilience against attacks such as paraphrasing and synonym substitution, while maintaining text-visual consistency, establishing new standards for quality-preserving multimodal watermarking