The semi annual reports for 2025 just released by various banks have updated the progress of large-scale model construction. The reporter found that in the first half of the year, the implementation of AI models in the banking industry showed an "accelerated running" trend, not only achieving rapid expansion of application scenarios, but also continuously breaking through and innovating in risk management scenarios. With the deepening of digital transformation, technological support capability has become the core cornerstone for banks to consolidate business development and layout future strategies, and many institutions have increased their investment in technology. Some banks have started planning their digital development blueprints for the next three to five years. The semi annual report on expanding the boundaries of application scenarios shows that commercial banks represented by state-owned banks are accelerating the landing of artificial intelligence applications and rapidly expanding the boundaries of application scenarios. Industrial and Commercial Bank of China (ICBC) disclosed in its 2024 annual report that it has established an enterprise level billion dollar financial model technology system called "ICBC Zhiyong", which empowers more than 20 major business areas and 200 scenarios. The semi annual report for 2025 shows that the bank has launched the "Leading AI+" initiative, adding more than 100 application scenarios such as AI wealth assistant and investment research intelligent assistant in key business areas such as personal finance, financial markets, and corporate credit. The Postal Savings Bank of China's "Postal Intelligence" big model adapts and absorbs multiple mainstream big models, and has carried out more than 230 big model scene constructions. The "You Win Together" interbank ecological platform integrates with the bill business trading robot "You Xiaoying" to achieve intelligent control of the entire process of all types of bills; The investment banking trading robot 'Youxiaobao' achieves intelligent inquiry response for bond underwriting transactions, improving inquiry transaction efficiency by over 95%; The "Intelligent Loan Review Assistant" supports over 30000 credit review scenarios such as agriculture, rural areas, and credit cards every day, effectively identifying, extracting, and classifying more than ten types of images, further improving loan review efficiency. Beijing Bank implements the "All in AI" concept and has established the "1213" AI system, which includes an integrated computing power base, two major model development and operation platforms, over 100 AI capabilities, and more than 300 AI application scenarios, to achieve coordinated management of models, computing power, requirements, and capabilities. Develop and launch a batch of AI tools such as "Xiaojing Caizhi" and chatBI. Retail business is the most suitable scenario for AI exploration and application. ”Wang Ying, Vice President of China Merchants Bank, stated at the performance briefing that the bank's AI construction and application mainly focus on "AI recruitment" and "AI assistance", and these applications have achieved certain results. At present, "AI Xiaozhao" serves over 20 million customers every month, and "AI Xiaozhu" has covered all positions in the bank. The reporter noticed that the application of large-scale models in the banking industry is evolving from efficiency improvement tools to value creation engines. With the continuous acceleration of the application of AI technologies such as big models in various fields of banking, we realize that 'AI+' will be a new opportunity for the industry and enterprises from the inside out. ”Vice President and Chief Information Officer of Postal Savings Bank of China, Niu Xinzhuang, stated. The information disclosed in the semi annual report on exploring intelligent risk management shows that the exploration of AI models in the field of risk management in the banking industry is currently mainly focused on risk investigation and early warning, anti fraud, anti money laundering and other scenarios. We adhere to the empowerment of digital intelligence, deepen the application of new technologies such as AI big models in the field of risk control, and continuously improve the level of intelligent risk control. The enterprise level intelligent risk control platform has been applied to all domestic branches and over 130 risk control decision-making scenarios, achieving intelligent risk investigation and early warning for the five major markets of commodities, foreign exchange, bonds, currencies, and stocks. ”Wang Jingwu, Vice President of Industrial and Commercial Bank of China, stated at the performance briefing that the bank has launched the industry's first credit AI intelligent agent matrix "Zhidai Tong", which achieves intelligent information capture, risk analysis and other functions. Develop an AI digital assistant for credit review called "Gongxiaoshen" to achieve rapid analysis and access to systems and data. Optimize the "Rong'an E-Defense" credit risk monitoring system, enrich statistical monitoring content, and improve data empowerment tools. Expand the coverage of satellite remote sensing monitoring scenarios, and enrich the application scenarios in agricultural fields such as planting and forest land. Zhejiang Commercial Bank disclosed in its semi annual report that it has iteratively upgraded its enterprise level risk management platform, including big data risk control and digital anti money laundering. It has established a "big model+small model" dual engine driven digital supervision system, focusing on improving and optimizing big data risk control models for general and characteristic scenarios. During the reporting period, more than 120 new business risk models were added, covering all business lines such as retail, supply chain, small and micro enterprises, and credit cards. In terms of fraud risk prevention and control, Postal Savings Bank of China relies on its big data platform base and data center capabilities, integrates AI and real-time stream computing, and constructs a full chain anti fraud model system. In the first half of 2025, more than 100000 potential victim accounts will be protected, preventing customer financial losses exceeding 800 million yuan. In terms of anti money laundering, the bank processes approximately 127 million transaction flow data daily, constructs nearly a hundred suspicious warning models, actively uses advanced technologies such as knowledge graphs and large models, and establishes a new type of anti money laundering suspicious transaction monitoring and reporting system with graph intelligence as the core. It achieves automated generation of suspicious analysis reports, effectively presents the evidence chain of gang money laundering, and helps improve manual identification efficiency by 30%. Industrial Bank also focuses on key areas such as anti money laundering and anti fraud, strengthening its digital security defense line. The semi annual report shows that in the first half of 2025, the bank intercepted 504 million yuan of fraudulent funds and protected 803 million yuan of transaction funds for potential victims. Industry insiders have pointed out that the application of AI big models in the field of risk management is still in the exploratory stage. In the future, it is necessary to gradually leverage the value of AI big models in risk identification foresight, risk assessment accuracy, risk monitoring speed, risk management breadth, and other aspects under the premise of safety and controllability. As the digital transformation enters the second half, banks are focusing on enhancing their technological support capabilities to lay a solid foundation for their future strategic layout. Bank of China continues to promote technological architecture transformation, with a total of 40000 cloud platform servers. Continuously upgrading the basic technology platform, as of the end of June 2025, the distributed technology platform has been integrated with 352 applications, the front-end technology platform has been promoted to 186 applications, and the big data technology platform has been promoted to 174 applications. Shanghai Pudong Development Bank has built an enterprise level intelligent foundation that integrates computing power, algorithms, platforms, and knowledge. As of the end of June 2025, we have built a thousand card computing power and taken the lead in implementing the financial application of "full stack localization computing power platform+open source big model" in joint-stock commercial banks, forming a "Puyin Zhiqi" model service matrix for large and small model collaboration. At the same time, we have established a multi-dimensional big model evaluation system and initially built a knowledge base for millions of enterprises. The distributed architecture system has basically taken shape, and the retail distributed core will smoothly achieve uninterrupted production of customer service in May 2025, providing effective support for the rapid development of the business. Building a panoramic blueprint for the construction of a "one cloud, multiple cores" financial cloud, forming a dual center deployment architecture, with a platform cloud host scale of over 35000 units and a container scale of over 275000 units. China Merchants Bank continues to strengthen its "cloud+AI+middle platform" technology foundation, establish a complete capability in large-scale model technology, and comprehensively promote the implementation of large-scale model applications. Cloud platform optimization and upgrading, focusing on consolidating availability and improving resource efficiency, continuously improving CMB Cloud. By the first half of 2025, the overall availability of the cloud will exceed 99.999%. In terms of technology platform, we focus on building industry-leading technical support capabilities, continuously improving component scale and quality. As of the end of June 2025, a total of 6314 components have been released. In addition, the bank is comprehensively promoting the construction and application of AI. At the technical system level, it is building an industry-leading intelligent computing infrastructure, continuously improving model performance and computing power efficiency through underlying technological innovation, creating an enterprise level AI middleware with low usage threshold and fast integration speed, and supporting the comprehensive promotion of large-scale model applications throughout the bank. At the same time, based on the trend of industry transformation and their own development reality, multiple banks have begun to plan their development blueprints for the next three to five years. Industrial Bank has also established a "Artificial Intelligence+" Action Leadership Group and Working Group, released the "Artificial Intelligence+Action Plan (2025-2027)", and comprehensively launched the "Artificial Intelligence+" action. Niu Xinzhuang revealed that Postal Savings Bank of China has launched the preparation of the "15th Five Year Plan" IT plan, which will lay out the blueprint for technological development, promote technological capabilities to a new level, and inject new momentum into the "Five Great Articles" of finance and promote high-quality development of finance. (New Society)