Research on Security Risks and Supervision of Generative AI Large Models

Generative AI Large Models represented by DeepSeek, ChatGPT,Midjourney and Sora have become the focus of discussions in the global tech circle once again. While the comprehensive rise of Generative AI has brought a lot of convenience to people, it has also brought a lot of risks. This paper studies the principles of large models and the differences from traditional ones, through analyzes the challenges or risks in multiple fields and dimensions such as ethical risks, security risks, social risks, legal risks brought by generative artificial intelligence, etc. By analyzing the regulatory agencies and mechanisms of different countries, targeted countermeasures and suggestions have been put forward for the supervision and governance in the future development process of generative artificial intelligence.