This paper introduces NeRFProtector, a plug-and-play watermarking framework for Neural Radiance Fields (NeRFs) to protect their copyright during creation. NeRFProtector uses a pre-trained watermarking base model to allow NeRF creators to embed binary messages directly while building their NeRFs. The framework ensures minimal modifications to the NeRF architecture, enabling flexible integration with various NeRF variants. A newly designed progressive distillation method is employed to efficiently embed and extract messages, achieving performance comparable to leading neural rendering methods. The proposed method leverages a watermarking base model, which can be sourced from a third party or trained separately. This base model is used to embed messages during NeRF creation, ensuring that messages are embedded before any potential misuse. The framework also introduces progressive global rendering (PGR), which renders all pixels across various resolutions, enabling more effective message embedding. This approach ensures that messages are robust and invisible, even under common image distortions. Experiments show that NeRFProtector achieves high bit accuracy and visual quality, outperforming existing methods in terms of consistency across different viewpoints. The method is robust against common image-level distortions and can be adapted to various NeRF variants. Additionally, the framework is effective against potential threats such as neural compression and adversarial attacks, demonstrating its resilience to malicious attempts to remove watermarks. The plug-and-play nature of NeRFProtector allows for easy integration with existing NeRF frameworks, making it a versatile solution for copyright protection. The method is compatible with different NeRF variants and can be applied to various scenarios, ensuring that the original data and NeRF models are securely managed. Overall, NeRFProtector provides an effective and efficient solution for protecting the intellectual property of NeRFs.This paper introduces NeRFProtector, a plug-and-play watermarking framework for Neural Radiance Fields (NeRFs) to protect their copyright during creation. NeRFProtector uses a pre-trained watermarking base model to allow NeRF creators to embed binary messages directly while building their NeRFs. The framework ensures minimal modifications to the NeRF architecture, enabling flexible integration with various NeRF variants. A newly designed progressive distillation method is employed to efficiently embed and extract messages, achieving performance comparable to leading neural rendering methods. The proposed method leverages a watermarking base model, which can be sourced from a third party or trained separately. This base model is used to embed messages during NeRF creation, ensuring that messages are embedded before any potential misuse. The framework also introduces progressive global rendering (PGR), which renders all pixels across various resolutions, enabling more effective message embedding. This approach ensures that messages are robust and invisible, even under common image distortions. Experiments show that NeRFProtector achieves high bit accuracy and visual quality, outperforming existing methods in terms of consistency across different viewpoints. The method is robust against common image-level distortions and can be adapted to various NeRF variants. Additionally, the framework is effective against potential threats such as neural compression and adversarial attacks, demonstrating its resilience to malicious attempts to remove watermarks. The plug-and-play nature of NeRFProtector allows for easy integration with existing NeRF frameworks, making it a versatile solution for copyright protection. The method is compatible with different NeRF variants and can be applied to various scenarios, ensuring that the original data and NeRF models are securely managed. Overall, NeRFProtector provides an effective and efficient solution for protecting the intellectual property of NeRFs.