360 Security Technology (SH:601360) Xin Lang Ke Ji·2025-08-07 05:14Core Insights - The president of 360 Digital Security Group, Hu Zhenquan, highlighted the complex AI security risks arising from both internal and external factors as AI technologies accelerate their deployment and application [1] Internal Risks - AI possesses inherent "four genetic defects": everything can be programmed, everything can be imitated, everything can be generated, and everything can be scheduled [1] - The complexity of AI security risks exceeds traditional categories, encompassing not only network and data security but also AI content security, application security, and future scenario security [1] External Risks - Attacks on AI systems are becoming a key aspect of geopolitical competition [1] - The black and gray markets are leveraging AI to mass-produce attack tools and deploy "hacker agents," which traditional security defenses struggle to counter [1] Proposed Solutions - 360 introduced a new approach called "mold to mold," utilizing AI technology to combat AI security risks [3] - The company developed the 360 Large Model Security Guardian, which consists of four security intelligent agents forming a closed-loop protection system [3] Security Intelligent Agents - The Content Security Agent acts as a gatekeeper for AI content, using specialized models for real-time content review to intercept illegal information and suppress false generation [3] - The AI Agent Security Agent focuses on execution safety, employing anomaly detection, permission control, and manual review to prevent data leaks and unauthorized actions [3] - The Software Security Agent scans for AI software vulnerabilities and configuration flaws, ensuring supply chain security from the source [3] - The Security Risk Assessment Agent simulates attacks to expose system weaknesses, helping clients identify critical vulnerabilities [3]