At a forum held in Beijing recently, Professor Robert H. Deng of Singapore Management University, Professor Song Xiaodong of the Department of electronic engineering and computing science of the University of California, Berkeley, and other scholars focused on the evolution direction of artificial intelligence technology, and proposed in their speeches that "there are security risks such as input attack and poison attack", and the formulation and implementation of artificial intelligence security specifications and standards is an urgent task. During the 2022 global digital economy conference artificial intelligence driven future industry forum, the topic of artificial intelligence safety and healthy sustainable development attracted wide attention. Robert H. Deng and song Xiaodong delivered speeches on topics such as "Research on development strategy based on artificial intelligence security" and "responsible, reliable and controllable AI security system". Zhang Jinsong, director of Beijing Municipal Bureau of economy and information technology, said in his speech that Beijing is exploring the establishment of a security risk management system for the integration of artificial intelligence applications, guiding the implementation of security risk assessment and improving the risk management capability. At the same time, it is accelerating the promotion of artificial intelligence data governance, strengthening the high-quality construction, high-level opening and high-level utilization of artificial intelligence data resources, and helping to build a Beijing artificial intelligence security ecosystem. In the view of Zhu Meng, partner and senior vice president of Ruilai intelligence, the construction of artificial intelligence security development specifications has ushered in a critical window period, and "safe and controllable" has become a basic capability that needs to be completed. She further proposed, including solving hidden dangers such as privacy leakage caused by strong dependence on data, improving the reliability of the algorithm, ensuring the stable operation of the system in case of intentional attacks, and the compliance and controllability of the application of deep synthesis technology. During the forum, Beijing artificial intelligence industry alliance and Jizhi future jointly released the "demonstration case of Beijing National artificial intelligence innovation and application pilot zone", which focused on the achievements of the construction of Beijing National artificial intelligence innovation and application pilot zone. Among them, realsafe, the artificial intelligence security platform of Ruilai intelligence, is also known as "anti-virus software for AI systems". Zhu Meng introduced that realsafe supports the detection of security risks against sample attacks and model backdoor attacks, and provides security scores, risk reports and various defense in depth solutions for the tested AI models. She also revealed that in view of the new security problems and industry needs in the field of artificial intelligence, as an AI enterprise initiated by the Institute of artificial intelligence of Tsinghua University, they are making new explorations in privacy computing, AI digital people, and "Ai face changing" detection with their own scientific research achievements and Industry landing experience. (news agency)