Commercial lenders shouldering heavier AI burden than central banks: ICBC president

As artificial intelligence transforms the finance sector, commercial bankers are grappling with a more challenging transition compared to central bankers, according to the president of Industrial and Commercial Bank of China (ICBC). Liu Jun, ICBC's leader, explained that while central bankers focus on oversight and policy, commercial lenders bear direct financial and legal responsibility when implementing AI at scale.

Speaking at a panel discussion at the Shanghai Bund Summit, Liu stated that when ICBC deploys AI tools across various business lines, he assumes responsibility for the deployment. Should something go wrong, he would be accountable to the victims and answer to the market. Commercial banks, which serve the public and industry daily, primarily rely on established technology, but they are just beginning their digital transformation journey, which remains uncertain.

Liu identified four key mismatches between AI and traditional banking: disclosure, data storage, sovereignty, and credibility. One major challenge lies in assigning liability when developers, deployers, and users each handle different aspects of an intricate system. ICBC has already deployed hundreds of AI models, granting them greater autonomy in data-intensive tasks while maintaining human teams for client interactions.

To address operational risks, ICBC has introduced a framework called the "proof of benefit," which directly links accountability to those who reap financial benefits and make independent decisions within the system. Arthur Yuen Kwok-hang, deputy chief executive of the Hong Kong Monetary Authority, echoed Liu's concerns about accountability as banks increasingly adopt agentic AI systems capable of planning and executing multi-step tasks.

Yuen explained that institutions often divide one application among six to eight agents, but if each agent and the human supervisor bear responsibility for only part of the task, "nobody is responsible for the whole thing." To tackle these challenges, Hong Kong's de facto central bank unveiled an updated AI governance framework emphasizing accountability, resilience, and trust last week.

This framework focuses on accountability tied to intent, a registry of material AI agents, and methods for intervention. Yuen expressed the Hong Kong government's aim to create an environment where financial institutions can safely "fail fast." He noted that supervisors will become more facilitative rather than regulatory, encouraging institutions to test their ideas.