Challenges and Remedies to Privacy and Security in AIGC: Exploring the Potential of Privacy Computing, Blockchain, and Beyond

Chuan Chen,Zhenpeng Wu,Yanyi Lai,Wenlin Ou,Tianchi Liao,Zibin Zheng

DOI: https://doi.org/10.48550/arXiv.2306.00419

2023-06-01

Cryptography and Security

Abstract:Artificial Intelligence Generated Content (AIGC) is one of the latest achievements in AI development. The content generated by related applications, such as text, images and audio, has sparked a heated discussion. Various derived AIGC applications are also gradually entering all walks of life, bringing unimaginable impact to people's daily lives. However, the rapid development of such generative tools has also raised concerns about privacy and security issues, and even copyright issues in AIGC. We note that advanced technologies such as blockchain and privacy computing can be combined with AIGC tools, but no work has yet been done to investigate their relevance and prospect in a systematic and detailed way. Therefore it is necessary to investigate how they can be used to protect the privacy and security of data in AIGC by fully exploring the aforementioned technologies. In this paper, we first systematically review the concept, classification and underlying technologies of AIGC. Then, we discuss the privacy and security challenges faced by AIGC from multiple perspectives and purposefully list the countermeasures that currently exist. We hope our survey will help researchers and industry to build a more secure and robust AIGC system.

What problem does this paper attempt to address?

The paper attempts to address the privacy and security challenges faced by Artificial Intelligence Generated Content (AIGC) during its rapid development. Specifically, the paper focuses on the following aspects:

1. **Privacy Leakage**: AIGC may leak user input data during content generation, especially during the data collection phase. Model developers might use unauthorized data for training, leading to data leakage issues.

2. **Model Attacks**: Malicious users can exploit vulnerabilities in AIGC models to conduct attacks. By adding interference signals to the input data, they can deceive the model into generating incorrect outputs or misleading information.

3. **Data Scraping**: Content generated by AIGC may be illegally scraped and misused, further raising data security concerns.

4. **Security of Generated Content**: AIGC-generated content may contain false or toxic information, which not only affects user perception but also involves ethical and moral issues.

5. **Copyright Issues**: AIGC-generated content may infringe on copyrights, especially when a large amount of existing data is used for training. Ensuring the originality and legality of the generated content is a significant issue.

To address these challenges, the paper systematically reviews the concepts, classifications, and underlying technologies of AIGC and analyzes the privacy and security issues faced by AIGC from multiple perspectives. Additionally, the paper explores the potential of advanced technologies such as blockchain and privacy computing in protecting AIGC data privacy and content security, proposing specific solutions and application cases. Through this research, the paper aims to help researchers and the industry build more secure and robust AIGC systems.