Artificial intelligence is like double-sided people, helping both good and bad people. Making good use of AI can create great functions, but if security management is not done well, AI can bring danger. Therefore, a very important aspect of the AI big model is security considerations. "said Zhou Dawei, Vice President of Technology for Asia Pacific and Japan at InBev. AI is like two sides of a coin, 'it can help both good people and tyranny', and behind this is the possibility of AI big models being 'poisoned'. Currently, the wave of AI big language models is sweeping the world, with domestic enterprises and universities launching self-developed big language models, and various industries are also exploring ways to empower AI. But AI can both "accelerate" efficiency and potentially become a tool for black and gray industries to make profits. In this context, what security risks will arise for the big model itself and the big model related industries? How should the domestic cybersecurity industry respond? Recently, reporters from Beijing News Shell Finance participated in the C3 Security Conference and the 2023 Global Digital Economy Conference Artificial Intelligence Summit Forum, and interviewed relevant experts. Some experts believe that the security problem of AI is not unique today, but in the wave of big models, it is necessary to prevent AI big models from being "polluted". At the same time, data security in the big model era will be more important than ever before. Whether the big model will lead to Data breach, and how to protect enterprise core technology assets are the key points of attention in the big model era. In order to prevent AI models from being "contaminated", Xu Yeli, Chief Security Officer of Asia CITIC Security, stated in an interview with a reporter from Shell Finance that the big language model technology is currently widely discussed, and it is indeed a revolutionary change. The AI industry has also reached the 2.0 era. In such a situation, there are more and more discussions about AI security. The relationship between AI and security includes: firstly, the AI system is getting larger, and its own security is a key factor. This involves AI "poisoning", model tampering, and the AI system itself has the ability to protect endpoints with boundary protection. Secondly, what should AI do if it does something bad? What should I do if AI may generate some misleading information? In addition, the extensive use of AI, especially overseas AI, may lead to data leakage, which is a significant risk and problem. Artificial intelligence is like double-sided people, helping both good and bad people. Making good use of AI can create great functions, but if security management is not done well, AI can bring danger. Therefore, a very important aspect of the AI big model is security considerations. "said Zhou Dawei, Vice President of Technology for Asia Pacific and Japan at InBev. Zhou Dawei stated that even with massive computing power, without a stable usage environment, content cannot be guaranteed against malicious attacks. There are two key points here. The first is that we need to be comprehensive and provide comprehensive protection, because not only do we store data, but we need to ensure data and security, ensuring that it is not attacked by others. We also need to ensure the transmission link and access link. Secondly, we need to be prepared in advance. In the era of artificial intelligence, we cannot use cat catching mouse methods, we need to deploy in advance and comprehensively ensure artificial intelligence We need to think more about the assets brought by the big model Professor in Chief of the Department of Computer Science at Tsinghua University, and Researcher in Artificial Intelligence at Tsinghua University