Technology Sharing

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

On July 3, the 2024 Global Digital Economy Conference "Cloud and Software Security Forum and the Second SecGo Cloud and Software Security Conference" was held in Beijing, hosted by the Global Digital Economy Conference Organizing Committee and co-organized by the China Academy of Information and Communications Technology (hereinafter referred to as CAICT) and the China Communications Standards Association. This forum focuses on the latest development trends of cloud and software security, and comprehensively demonstrates CAICT's research and exploration in the field of innovative security through topics such as cloud security, zero trust, security big models, and software supply chain governance. The fourth batch of "Security Guardian Program" outstanding cases were announced at the conference. Baidu's security operation intelligent body platform, with its leading technological advantages and significant industry application value, was successfully selected as the "2024 Security Guardian Program Security Big Model Special Outstanding Case" after multiple reviews. During the meeting, the "Security Industry Big Model Technology Application Trend Development Report" co-written by Baidu Security was also officially released at the conference.

2024 Security Guardian Program Security Model Special Topics Excellent Cases

As the digitalization and informatization process of enterprises accelerates and cybersecurity threats increase, in order to effectively protect information assets and ensure network and data security, many organizations have begun to adopt large-scale artificial intelligence models for security defense and response. Through artificial intelligence models, massive amounts of data are analyzed to identify and predict potential security threats, provide immediate alerts and responses, detect abnormal behaviors, identify malware, predict attack trends, and help organizations take timely measures to respond to various security risks. In order to guide the development direction of products in the security field, China Academy of Information and Communications Technology launched the "Security Guardian Program | Security Big Model Special Topic" excellent case collection activity to further promote the in-depth application of security technology by setting benchmarks. After several months of rigorous selection and review, Baidu's security operation intelligent body platform has received unanimous praise from judges, experts and users, and won the 2024 Security Guardian Program Security Big Model Special Topic Excellent Case.

Baidu's security operation intelligent body platform has a full-process intelligent security operation solution, and adopts a layered technical architecture to continuously optimize the business effects of intelligent operation scenarios. It is divided into the model layer, framework layer and application layer from the bottom up, providing intelligent capabilities for key security operations, including threat intelligence analysis, intrusion detection, alarm analysis, deep source tracing analysis, intelligent stop loss, security reinforcement, event report generation, etc., effectively improving operational efficiency and security protection effects in all aspects of security operations.

Baidu Security Operations Intelligent Platform Architecture

Baidu's security operation intelligent agent platform has technical innovations aimed at practical effects at every level. At the data and model level, it innovatively combines the thinking chain capabilities of the big model and the security TTP framework to build data, and adopts multiple training methods such as Post-Pretrain, SFT, and Dagger to jointly improve the basic capabilities of the big model; at the architecture level, it combines the characteristics of the security field to build an expert-agent collaboration framework, builds a management method that supports multi-source heterogeneous knowledge, and builds a security tool set to support actual operation tasks; at the application level, it adopts dynamic and static combined workflows, scene knowledge embedding, and task-level memory to effectively improve the effectiveness of intelligent agent tasks; at the application level, it adopts a design that combines LUI and GUI to make it easier for security experts to control the entire security operation process and interact with Agents. Baidu's security operation intelligent agent platform significantly improves the efficiency and effectiveness of each link such as intrusion detection, alarm analysis, event triage, attack tracing, and report generation. The platform can analyze massive data and summarize information, accurately explore potential security risks, and strengthen the overall security situation. By reducing the number of tedious tasks for security experts, they can focus on core operations. At the same time, the platform also effectively utilizes the experts' handling experience, enhances the team's professional knowledge, and comprehensively improves security operation capabilities.

At the same time, the "Development Report on the Application Trends of Big Model Technology in the Security Industry (2024)" released at the meeting further sorted out the application status of big models in the security industry, deeply analyzed the key to the implementation of big model technology in my country's current security industry, and systematically summarized the main application scenarios of big models in the security field, in order to explore the new trend of the integration of traditional security technology and big models, and gain insight into the future development direction of the security industry in the era of artificial intelligence.

"Development Report on the Application Trend of Big Model Technology in the Security Industry (2024)"

In the future, Baidu Security will continue to cooperate with partners from all walks of life in industry, academia and research to jointly meet the new opportunities and challenges of the big model era, work together to build a security line of defense, reconstruct security business processes through intelligent collaboration, build a new quality of productivity of AI in security big models, and help security practitioners protect thousands of industries.