ProMark: Proactive Diffusion Watermarking for Causal Attribution
This addresses the issue of recognition and reward for creatives whose content is used in generative AI training, though it is an incremental improvement in attribution methods.
The paper tackles the problem of attributing synthetically generated images to their training data concepts by proposing ProMark, a proactive diffusion watermarking technique that embeds imperceptible watermarks into training images and retains them in generated images, achieving the ability to embed up to 2^16 unique watermarks while maintaining image quality and outperforming correlation-based attribution.
Generative AI (GenAI) is transforming creative workflows through the capability to synthesize and manipulate images via high-level prompts. Yet creatives are not well supported to receive recognition or reward for the use of their content in GenAI training. To this end, we propose ProMark, a causal attribution technique to attribute a synthetically generated image to its training data concepts like objects, motifs, templates, artists, or styles. The concept information is proactively embedded into the input training images using imperceptible watermarks, and the diffusion models (unconditional or conditional) are trained to retain the corresponding watermarks in generated images. We show that we can embed as many as $2^{16}$ unique watermarks into the training data, and each training image can contain more than one watermark. ProMark can maintain image quality whilst outperforming correlation-based attribution. Finally, several qualitative examples are presented, providing the confidence that the presence of the watermark conveys a causative relationship between training data and synthetic images.