At the 2nd Cyber Space Security (Tianjin) Forum held on September 4, 2024, Zhou Hongyi, founder of 360 Group, delivered a speech titled "Building a Secure Large Model to Reshape Security with AI," sharing his insights on the cybersecurity landscape in the era of AI large models and introducing 360 Group's practices and achievements in this field.
Zhou Hongyi revealed that 360 Group has invested nearly 30 billion RMB in cybersecurity over the past decade, establishing a robust computing infrastructure including 210 data centers and 200,000 servers, and assembling Asia's largest team of senior experts. He emphasized that despite the significant investment, 360 Group will continue to increase its investment to support national cybersecurity.
In his speech, Zhou Hongyi pointed out that AI large models are reshaping all industries and products. 360 Group is exploring the use of AI technology to enhance security effectiveness and address the pain points in the cybersecurity industry. He believes that through large model empowerment, challenges such as massive data analysis, rapid response, and traceability can be quickly resolved. At the same time, he also warned that the rapid development of AI technology brings new security challenges, including AI attack and defense security, AI self-security, and AI ethical security issues.
Zhou Hongyi proposed that 360 Group's attack detection expert model, developed based on artificial intelligence technology, has transitioned from "rule-driven" to "learning-driven," with the ability to detect "known unknowns" and "unknown unknowns." He compared 360's security large model to having L4 "self-driving" capabilities and considered this the key to achieving automated cybersecurity operations.
He also emphasized that the key to building a secure large model lies in rich large model experience, training data, and deep integration with security business. In this process, 360 Group has built multiple expert models including attack detection, operational response, traceability, knowledge management, data protection, code security, and content security, and has surpassed GPT-4 in security capabilities.
Finally, Zhou Hongyi announced that 360 Group will provide large model standard capabilities to all users who purchase its security products and will carry out security large model empowerment upgrades in the future to enhance new quality productivity and combat effectiveness in security.