Core Viewpoint - The Beijing Economic and Technological Development Zone Management Committee has issued an implementation plan for accelerating the construction of a comprehensive artificial intelligence city from 2026 to 2027, focusing on enhancing foundational capabilities in AI and integrated circuit manufacturing [1] Group 1: AI Infrastructure Development - The plan emphasizes leveraging integrated circuit manufacturing capabilities to promote a collaborative development model that integrates design, manufacturing, testing, and computing power [1] - It aims to drive continuous iteration of computing infrastructure and high-performance intelligent computing industries [1] Group 2: Chip Architecture Innovation - The initiative includes advancing research and innovation in chip architectures such as mixed-signal and storage-computing integration [1] - It seeks to extend the industrial chain to include scenario-defined chips, industry-specific chips, and enabling software [1] Group 3: Model Development and Testing - The plan supports the collaborative development of general and vertical models, accelerating the upgrade from cognitive models to proactive models [1] - It also promotes the establishment of national-level AI hardware and software testing verification centers and large model evaluation centers to strengthen foundational capabilities in "chip-model adaptation" [1] Group 4: AI Security Framework - The initiative actively deploys an AI security system, focusing on breakthroughs in deep forgery detection and compliance checks for generated content [1] - It aims to build a secure and trustworthy foundation for AI applications [1]

Core Insights - AI security has become a critical foundation for high-quality industrial development as the "AI+" initiative accelerates [1][2] - The forum emphasized the need for a collaborative ecosystem to establish standards, assessments, and protective frameworks for AI security [1][4] Group 1: AI Security Governance - The forum gathered various stakeholders to discuss cutting-edge issues in AI security governance and released multiple research outcomes and industry standards [1][2] - Recommendations were made to strengthen the technological foundation, deepen application integration, and improve governance ecosystems for AI security [1][2] Group 2: Policy and Industry Development - The "14th Five-Year Plan" emphasizes enhancing national security capabilities in emerging fields like AI, guiding future work in the information and communication sector [2] - The AI security industry in China is entering a phase of high-quality development, with continuous optimization of the policy environment and ongoing technological innovations [2][3] Group 3: AI Safety Challenges and Solutions - The rapid evolution of large models and intelligent agents has led to new risks such as identity fraud and decision-making failures, necessitating comprehensive safety measures [3][4] - A dual authorization mechanism for user and application interactions was proposed to mitigate privacy and data leakage risks in cloud-based intelligent agents [3][4] Group 4: Industry Collaboration and Standards - The forum initiated the development of the "AI Native Cloud Security Capability Maturity Requirements" standard to provide a quantifiable guide for building AI-native cloud security [5] - Experts from various companies discussed the challenges and solutions in AI security, emphasizing the need for an open and collaborative industry ecosystem [5]

Core Viewpoint - The article emphasizes the rapid evolution of large models in artificial intelligence (AI) and their significance as indicators of national innovation capability and comprehensive national strength. It highlights the security challenges posed by these models, particularly in the context of national security and public interest [2][3]. Group 1: Current State of AI Models - As of January 2025, there are approximately 197 large models in the Chinese market, covering key industries such as finance, healthcare, education, manufacturing, automotive, and energy [2]. - Global large models face security issues, including the generation of false/harmful content, data bias, and information leakage, which pose significant threats to national information security [2]. Group 2: Security Challenges and Responses - Domestic platforms face challenges in framework completeness, developer community maturity, and open-source ecosystem development, with some early versions of domestic large models showing a jailbreak failure rate of up to 100% [3]. - Zhejiang University and Huawei have launched the DeepSeek-R1-Safe foundational model, which has improved security defense capabilities to 83%, a 115% increase compared to the original model [3][5]. Group 3: Technical Innovations - DeepSeek-R1-Safe incorporates breakthroughs in three dimensions: "secure corpus construction," "secure model training," and "hardware and software environment setup" [4][5]. - The model's training process is fully deployed on the domestic Ascend Kunpeng cluster, utilizing 128 servers and a total of 1024 Ascend AI cards, marking a significant achievement in large-scale security training [9][10]. Group 4: Performance Metrics - DeepSeek-R1-Safe demonstrates nearly 100% success in defending against ordinary harmful issues across 14 dimensions, outperforming several contemporaneous models by 4% to 13% [10][12]. - The model's jailbreak defense capability exceeds 40% against various jailbreak modes, surpassing contemporaneous models by 16% to 23% [13][15]. Group 5: Future Directions - The team aims to promote the development of endogenous secure AI in collaboration with Huawei and other industry partners, focusing on achieving comprehensive autonomy, security, and controllability in AI models [18].