As artificial intelligence (AI) reshapes economies and institutions worldwide, central banks are beginning to grapple with how to harness its potential without compromising stability. Speaking at the Global Fintech Fest 2025, Tao Zhang, Chief Representative, Bank for International Settlements (BIS) outlined how AI is redefining central banking operations — from economic forecasting to risk management — while also underscoring the pressing need for governance, transparency, and ethical oversight.

Zhang began by noting that central banks today operate in an “information-driven world” that demands not just financial prudence but also technological fluency. “AI can support operations, develop talent, and enhance strategies for central banks, helping the world navigate financial and economic challenges,” he said.

AI as a Tool for Economic Insight

Zhang highlighted how central banks across countries are already using AI to improve decision-making.
Institutions such as the Bank of Canada and the Reserve Bank of India have begun deploying AI models to monitor economic activity by analysing real-time data and social media trends. Similarly, the Federal Reserve Bank of New York uses machine learning tools to assess consumption patterns, supply chain bottlenecks, and property market shifts, offering policymakers more timely and accurate insights.

“AI models can process massive datasets, identify patterns in financial statements, and even detect anomalies during times of financial stress,” Zhang said. Open-source large language models (LLMs), he added, can extract economic narratives and reveal relationships that traditional tools might miss.

Driving Efficiency in Payments and Compliance

Beyond analytics, AI is also transforming how central banks and financial institutions manage payments and compliance. Zhang pointed to how AI-powered systems are improving Know Your Customer (KYC) processes, detecting unusual transactions, and reducing operational costs. Central banks in Japan and Korea, for instance, are investing in AI projects to engage the next generation of economists and policymakers.

Challenges: From Hallucinations to Data Bias

Yet, as Zhang cautioned, the road ahead is not without challenges.
AI’s reliability remains a concern, especially when large language models “hallucinate” — producing plausible yet inaccurate information. “Open-source models may lack the robustness required for critical applications in monetary policy or financial stability,” he warned.

Data privacy, cybersecurity, and regulatory compliance are equally pressing concerns. “Central banks must balance the speed and flexibility of AI-driven insights with the need for careful, responsible decision-making,” he said, citing recent findings from the European Central Bank and Italian authorities.

Biases embedded in training data can also lead to unfair outcomes in areas like credit assessment or financial product allocation. Zhang emphasised that human oversight remains indispensable: “AI should support—but not replace—human judgment.”

The BIS Approach: Responsible Experimentation

To guide central banks through this transition, BIS is focusing on three core priorities:

1. Knowledge sharing and collaboration among central banks and economists.
2. Responsible experimentation with AI tools that complement, not supplant, traditional policy mechanisms.
3. Advancing research and development to strengthen economic monitoring, risk analysis, and financial stability.

A Call for Balance

Concluding his remarks, Zhang said AI represents a “transformative force” for the financial system — one that can enhance efficiency, improve inclusion, and strengthen policy frameworks if used responsibly.
“The challenge,” he said, “is to strike a balance between embracing technological innovation and preserving the integrity, stability, and fairness of the financial system.”