Technological and industrial revolution are figuratively emerging into prominence from obscurity. As one of driving forces leading emerging technological and industrial changes in the world, artificial intelligence (AI) technology will accelerate its transformation into infrastructure, enabling the whole industry and society. However, potential security threats and risks in AI infrastructure have been recognized. The endogenous security of AI (AI Security) is one of the most fundamental component to the transformation. In recent years, both academia and industry have been strengthening the deployment and cutting-edge research of AI Security.
In 2019, Ant Group and Zhejiang University established the “ Zhejiang University – Ant Group Fintech Center ”. The “ Date Security and Privacy Protection Laboratory ” is the first laboratory established in the center. Professor Kui Ren of Zhejiang University and Tao Wei, Vice President of Ant Group, served as co-directors. The “ Artificial Intelligence Security White Paper (2020) ”, one of the results of this lab, took nearly a year to be researched and written. The full version of the White Paper is nearly 60,000 words. This White Paper comprehensively elaborates AI security from the common perspective of academia and industry, systematically summarizes the security threats faced by AI models, AI data and AI carrying systems. What’s more, according to different application scenarios and threat models of attackers, the advantages and disadvantages of relevant researches in recent years are analyzed. In addition, the further development trend of AI security technology is discussed, and a systematic security solution to AI application is proposed for the future.
AI Security still has a long way to go. Zhejiang University and Ant Group expect to contribute to Artificial Intelligence Security together with colleagues from all walks of life in the future. Through the development of provably safe artificial intelligence, a solid security foundation will be laid for the national AI strategy and strong guarantees will be provided for the development of AI. It will make AI technology truly create new and reliable productivity in people’s daily lives.
Artificial Intelligence Security White Paper(2020)
Data Security and Privacy Protection Laboratory, Zhejiang University-Ant Group Fintech Center December, 2020