The field of watermarking for AI-generated content is rapidly evolving, with a focus on developing innovative methods to detect and prevent copyright infringement. Researchers are exploring new approaches to watermarking, including hardware-based authentication and fragile watermarking using deep steganographic embedding. These methods aim to provide a more secure and reliable way to verify the authenticity of AI-generated content, particularly in the face of increasingly sophisticated generative models. Notably, some papers are proposing novel two-step image generation models and diffusion model watermarking methods that address the challenges of copyright protection and inappropriate content generation. Overall, the field is moving towards more robust and practical solutions for watermarking AI-generated content. Noteworthy papers include:

• On-Device Watermarking, which argues for a socio-technical framework for watermarking via cryptographic signatures,
• Gaussian Shading++, a diffusion model watermarking method tailored for real-world deployment,
• VideoMark, a training-free robust watermarking framework for video diffusion models.