🤖 AI Summary
To address the high retraining cost of watermarking methods under frequent LDM iterations in the AIGC era, this paper proposes a training-free, plug-and-play latent-space watermarking framework for Stable Diffusion (SD) models to enable copyright tracing and secure content governance. Methodologically, it dynamically embeds watermarks into the latent space by modeling the diffusion process, employing gradient-free optimization and multi-attack robustness enhancement—without modifying model architecture. It achieves, for the first time, zero-shot cross-version compatibility across SD 1.5, 2.1, and XL. The embedded watermark is imperceptible and preserves image fidelity. Experiments show a watermark detection accuracy exceeding 99.2%; even after JPEG compression, cropping, and filtering attacks, detection rates remain above 92%, significantly outperforming existing training-dependent approaches.
📝 Abstract
Nowadays, the family of Stable Diffusion (SD) models has gained prominence for its high quality outputs and scalability. This has also raised security concerns on social media, as malicious users can create and disseminate harmful content. Existing approaches involve training components or entire SDs to embed a watermark in generated images for traceability and responsibility attribution. However, in the era of AI-generated content (AIGC), the rapid iteration of SDs renders retraining with watermark models costly. To address this, we propose a training-free plug-and-play watermark framework for SDs. Without modifying any components of SDs, we embed diverse watermarks in the latent space, adapting to the denoising process. Our experimental findings reveal that our method effectively harmonizes image quality and watermark invisibility. Furthermore, it performs robustly under various attacks. We also have validated that our method is generalized to multiple versions of SDs, even without retraining the watermark model.