Copyright Protection
Copyright protection in the age of generative AI focuses on safeguarding intellectual property rights in the context of models that create novel content. Current research emphasizes developing robust watermarking techniques for various model architectures, including diffusion models, neural radiance fields (NeRFs), and large language models (LLMs), often employing methods like backdoor watermarking and adversarial examples. These efforts aim to address the challenges posed by the ease with which generative AI can reproduce copyrighted material, impacting both legal frameworks and the practical application of these powerful technologies.
Papers
OmniGuard: Hybrid Manipulation Localization via Augmented Versatile Deep Image Watermarking
Xuanyu Zhang, Zecheng Tang, Zhipei Xu, Runyi Li, Youmin Xu, Bin Chen, Feng Gao, Jian Zhang
CopyrightShield: Spatial Similarity Guided Backdoor Defense against Copyright Infringement in Diffusion Models
Zhixiang Guo, Siyuan Liang, Aishan Liu, Dacheng Tao