Core Insights - The 2025 Global Digital Economy Conference highlighted the current state of AI model security in China, indicating that it is mostly at a mid-level with an attack success rate of approximately 5%-6%, suggesting room for improvement [1][3]. Group 1: AI Security Challenges - Experts noted that AI is amplifying security threats, making them more scalable and targeted, which exacerbates the asymmetry in cyber defense [3]. - There is a significant gap between China's AI security algorithms and international standards, particularly in originality, model openness, and dataset quality [3]. - A new security assessment platform for AI models revealed that advancements in AI capabilities have not corresponded with improvements in security risk prevention [3]. Group 2: Importance of Systematic Defense - Experts emphasized the need for a comprehensive network security system, moving away from isolated defense strategies to a unified defense framework [4]. - The successful zero-incident network protection during the 2022 Winter Olympics was cited as an example of the effectiveness of a joint defense system [4]. - Recommendations for establishing a joint defense system include ensuring operators fulfill their security responsibilities and transforming individual service purchases into a public service under a unified defense framework [4]. Group 3: Industry Collaboration and Governance - The importance of collaborative principles in industry chain integration was stressed, advocating for deeper integration of technology and international standards [5]. - A shift from individual product selection to a systematic approach in security construction was proposed, addressing the fragmentation in security procurement among large enterprises [6]. - Suggestions for enhancing security operations include leveraging AI for threat intelligence and establishing efficient feedback loops for security operations [6].

Core Viewpoint - The cybersecurity industry is facing significant challenges due to a fragmented system, where the procurement of security devices from multiple vendors leads to dispersed responsibilities and ineffective security capabilities [1][2] Group 1: Systemic Challenges - Over 90% of large government and enterprise organizations procure security devices from more than 10 vendors, resulting in a "big platter" of fragmented security systems [2] - There is a disconnect between the macro-level "big system" and the micro-level "small systems," leading to difficulties in integrating security capabilities and implementing unified strategies [2][3] - The operational bottleneck arises from the lack of data sharing and communication between different "small systems," which exacerbates the security risks [2] Group 2: Impact of AI - The traditional strategy of protecting core "small data" through compartmentalization has become ineffective in the age of AI [3][4] - Small data, which includes trade secrets and proprietary information, is now at risk of being easily accessed and replicated by competitors using AI technologies [4] Group 3: Proposed Solutions - To address these challenges, the industry must focus on three key paths: reshaping data aggregation models, constructing feedback loops for security operations, and redefining ecological cooperation models [5] - A comprehensive security data set that covers a wide range of dimensions and has high contextual relevance is essential for effective threat detection [5] - Establishing a "security system chief designer" to unify disparate security efforts and create a cohesive defense strategy is crucial for effective implementation [6]