Financial sector cautious on emerging tech adoption due to security concerns

Two large State-owned commercial banks in China have announced the rollout of their own versions of OpenClaw, signaling growing interest in artificial intelligence across the sector.

The push, however, is not without some hesitation. Analysts said that as highly regulated and risk-averse institutions, lenders cannot directly deploy open-source AI agents. Instead, they must rely on private deployment and enhanced security safeguards — placing high requirements on their capabilities in risk management, technology and talent.

Recently, Postal Savings Bank of China collaborated with openJiuwen — an open-source agent framework and platform — to launch the PSBCClaw ecosystem. As a major breakthrough in the bank's AI agent development, PSBC-Claw is a customized solution based on JiuwenClaw, an intelligent AI agent built on openJiuwen.

READ MORE: Beijing researchers release safety detection tool for OpenClaw

Featuring core advantages such as autonomous control, trusted authorization, robust security, centralized management and multimodal capabilities, PSBC-Claw is supported by large-model clusters and is expected to reshape the paradigm of intelligent financial services.

To meet the financial sector's stringent requirements for compliance, security, risk control and minimal authorization, PSBC-Claw has comprehensively upgraded its security framework. The system has established an end-to-end security control mechanism covering data access, knowledge updates, skill authorization, model computation and results output, effectively ensuring authentication, business operations, behavior control, data protection and environmental security for financial AI agents during task execution.

PSBC-Claw has significantly improved the efficiency of real-time intelligence monitoring, enabling minute-level, around-the-clock precise acquisition of information, intelligent analysis and real-time multichannel push notifications. It has already been applied internally at PSBC in areas such as intelligence monitoring, risk early warning and technical insights.

The system operates 24/7 with full automation, efficiently executing end-to-end monitoring tasks while continuously optimizing monitoring rules and analytical models to improve accuracy, helping users gain precise insights into market dynamics. In the future, PSBC plans to gradually expand the use of PSBC-Claw to areas such as administration, operations, credit services and risk management.

In March, Gu Shu, chairman of Agricultural Bank of China, also revealed that the bank had launched its own AI agent, ABCClaw. The system supports relationship managers by automatically processing data on green projects, cross-validating multidimensional information and intelligently generating due diligence reports, thereby making the lending process more convenient, efficient and secure.

Despite these developments, the Chinese banking sector still remains cautious in its adoption of Open-Claw. Tian Lihui, a professor of finance at Nankai University, said that Open-Claw's default high system privileges and relatively weak security configurations make it vulnerable to exploitation by hackers, potentially becoming an entry point for data breaches or unauthorized transaction manipulation — posing a clear conflict with banks' stringent security and compliance requirements.

Industry experts also pointed out that Open-Claw requires large volumes of data for training and optimization. Achieving higher levels of intelligence often requires users to provide more data and potentially sensitive personal information. For banks — which manage vast databases and are closely tied to financial security — this presents significant challenges.

ALSO READ: Cloud providers seen pushing up AI service rates

At the 2026 science and technology work conference of the People's Bank of China — the country's central bank — on March 11, authorities explicitly called for deeper integration of business and technology, and for the active, prudent, safe and orderly advancement of AI applications in the financial sector to unleash momentum for digital and intelligent development.

Lou Feipeng, a researcher at Postal Savings Bank of China, suggested that banks should adopt a prudent approach to the deployment of AI agents — starting with small-scale pilots focused on low-risk scenarios, and gradually expanding after validating their effectiveness. Lou also stressed the importance of building an end-to-end data security system, employing desensitization and encryption technologies, and clearly defining the boundaries for data usage.

Contact the writers at jiangxueqing@chinadaily.com.cn