Artificial Intelligence (AI) and Machine Learning (ML) are transforming the global financial landscape by enabling systems to analyze complex data, predict outcomes, and automate decision-making. While AI broadly refers to machines simulating human intelligence, ML is a subset where algorithms learn patterns from data to improve predictions and insights over time. Across the globe, central banks are increasingly harnessing Generative AI (GenAI) and Large Language Models (LLMs) to enhance decision-making, supporting faster data analysis, deeper insights, and improved monitoring mechanisms. Increasingly, AI is also being integrated into central banks’ market intelligence (MI) activities, enhancing how monetary authorities gather, interpret, and fill gaps in real-time market data. Tools such as natural language processing (NLP) and sentiment analysis are deployed to capture unstructured data, providing sharper insights into economic expectations and policy impacts (BIS, 2025). These developments raise challenges around data confidentiality, algorithmic bias, and skill shortages, underscoring the need for robust governance frameworks (BCBS, 2025). AI applications span a wide range of use cases, including inflation forecasting, anomaly detection, sentiment analysis, regulatory data management, anti–money laundering (AML), tokenization experiments, and central bank digital currency (CBDC) testing. While approaches differ across jurisdictions, there is a shared focus on responsible innovation—leveraging technology to modernize financial systems without compromising stability, transparency, or trust.

It is recommended that central banks integrate AI into stress testing and macroprudential oversight to mitigate these risks. It also highlights the role of central bank digital currencies (CBDCs), where AI optimizes transaction security and fraud detection, with China’s digital yuan as a case study (IMF, 2025). Yet, unequal AI access across countries—only 20% of low-income nations have AI-ready infrastructure—challenges global financial coordination. The capacity-building initiatives and knowledge sharing are crucial to bridge this gap. For central banks, adopting AI requires balancing innovation with ethical considerations, such as addressing biases in credit scoring algorithms that could exacerbate inequality. It is therefore significant for the central banks to evolve their mandates to address AI’s macroeconomic and social implications, ensuring inclusive growth and stability. This blog presents succinctly the cross-country experiences on AI in central banking.

Country-Level Developments of AI in Central Banking

USA: US central bank has adopted machine learning (ML) for real-time economic projections. This experiment at US Fed Reserve points to the fact that AI-enhanced forecasting models can improve inflation targeting accuracy. However, as employment is the basic mandate of central banks in the advanced economies, labour market disruptions by AI is a matter of urgent concern. In the advanced economies, 40% of jobs are at risk of automation, necessitating proactive central bank policies to stabilize demand (IMF, 2025). AI also introduces financial stability risks, such as algorithmic trading amplifying market volatility, as evidenced by a 2024 flash crash linked to AI-driven trades.

Euro Zone: The European Central Bank (ECB) is leveraging an innovation sandbox to test programmable payments, tokenization, and other advanced use cases as part of its digital euro project. To further support the development of the digital euro, the ECB has launched the CBDC Innovation Platform, a collaborative testing environment enabling financial institutions, technology providers, and regulators to experiment with potential applications. In addition, the ECB leverages advanced media analytics, processing over 6 million pages annually, machine-translated across 24 EU languages, to monitor economic sentiment, assess policy impacts, and identify emerging risks.

Germany: The Deutsche Bundesbank has developed MILA (Monetary-Intelligent Language Agent), an AI-powered model designed to analyse central bank communication with precision and transparency. Using advanced language models, MILA evaluates monetary policy statements sentence by sentence, classifying them as hawkish or dovish while considering broader economic context. It also analyses tone—positive, neutral, or negative—and provides justifications for its classifications, enabling experts to validate results. MILA is used to assess ECB press conferences, policy statements, and speeches, offering insights into the impact of communication on markets.

United Kingdom: The Bank of England (BoE) is piloting generative AI tools and LLMs for internal use, including drafting analyses, summarizing reports, and improving inflation forecasting. Complementing this, the UK’s Financial Conduct Authority (FCA) has launched a comprehensive AI Innovation Lab, featuring AI Spotlight, AI Sprint, AI Input Zone, and the Supercharged Sandbox, offering enhanced datasets and testing capabilities.

Singapore: The Monetary Authority of Singapore (MAS) has advanced Project MindForge, partnering with banks, tech firms, and industry associations to develop a risk framework and reference architecture for GenAI use in finance. It is also preparing an AI governance handbook and exploring use cases in sustainability and cybersecurity. MAS has introduced the Pathfinder Programme to accelerate adoption, curating a library of AI use cases and best practices while partnering with training providers for skills development. As of 2025, 20 financial institutions across banking, insurance, and payments have joined the initiative.

India: The Reserve Bank of India (RBI) has introduced the FREE-AI framework—Framework for Responsible and Ethical Enablement of AI—to guide safe and effective adoption across the financial sector. RBI reports suggest that generative AI could improve banking efficiency by up to 46% (EY estimates), enabling advancements in customer service, fraud detection, compliance, and inclusion. A key initiative is an AI Innovation Sandbox to enable experimentation and technical readiness assessment. Unlike a traditional sandbox, this platform offers infrastructural support but no regulatory relaxations, running alongside RBI’s existing Regulatory Sandbox where AI-driven solutions may still access regulatory flexibilities.

Australia: The Reserve Bank of Australia (RBA) has taken a multi-faceted approach to AI. Its RBA PubChat, an AI-powered chatbot, analyzes nearly 20,000 analytical documents spanning four decades to track sentiment, detect signals on price pressures, and generate aggregated indicators. In parallel, the RBA has advanced CBDC research, including a pilot project with the Digital Finance Cooperative Research Centre and Project Acacia, exploring tokenization and distributed ledger technology (DLT) to enable wholesale tokenized asset markets.

Canada: The Federal Reserve is exploring AI and LLMs for economic forecasting, financial stability monitoring, and supervisory oversight. The Bank of Canada is adopting AI for inflation nowcasting, stress testing, and sentiment analysis. Unlike other jurisdictions, the U.S. and Canada central banks lack a national regulatory sandbox, though several individual states have established their own frameworks.

Middle East: The Central Bank of the United Arab Emirates (CBUAE) has partnered with Presight to support the Financial Infrastructure Transformation programme, deploying sovereign AI-powered platforms for critical systems including CBDCs, instant payments, card schemes, and RTGS. The Saudi Central Bank (SAMA) operates a Regulatory Sandbox aligned with Vision 2030, enabling financial institutions and fintechs to experiment with AI-powered solutions in payments, compliance, and lending.

Hong Kong: The Hong Kong Monetary Authority (HKMA), in collaboration with the Hong Kong Cyberport Management Company, launched a Generative AI Sandbox in 2024. This provides a risk-managed framework with technical support and supervisory feedback, allowing banks to safely pilot GenAI use cases.

China: The People’s Bank of China (PBoC) has highlighted AI’s role in security, innovation, and data governance, while advancing the digital yuan (e-CNY). It has also issued guidelines to regulate AI-based financial applications and strengthen data security.

Comparative Overview: Central Banks’ AI and GenAI Initiatives

While each central bank has adopted a different approach to AI, their strategies converge on a few common themes—enhancing forecasting, improving policy communication, and strengthening supervisory capabilities. The following table highlights initiatives from leading central banks that have published official papers, reports, or frameworks on the use of Artificial Intelligence (AI) in monetary policy, financial supervision, operational efficiency, use cases.

As the table 1 illustrates, while central banks differ in their specific initiatives, there is a strong convergence in priorities: adopting AI to enhance decision-making, improving operational efficiency, and managing risks effectively.

Global Collaboration

At the cross-border level, the Bank for International Settlements (BIS) has launched Project Aurora through its Innovation Hub Nordic Centre to strengthen anti–money laundering (AML) frameworks. Using AI, ML, and privacy-enhancing technologies, Aurora applies a data-driven, behaviour-based approach to detect suspicious transactions, hidden networks, and anomalous patterns while ensuring confidentiality through collaborative learning. Building on these efforts, the IMF’s 2025 survey of 191 central banks highlights both opportunities and risks. AI could boost global GDP by 7% by 2030, but uneven adoption—especially since only 20% of low-income countries have AI-ready infrastructure—risks widening disparities. For central banks, this implies adapting monetary and macroprudential frameworks, as AI-driven productivity in advanced economies may stoke inflation while lagging adoption in LICs threatens financial stability and inequality.

Balancing Innovation with Risk While AI adoption in central banking is accelerating, central banks are careful not to over-rely on technology. Their approaches are grounded in: transparency and explainability to ensure AI-driven insights are interpretable; ethical governance to prevent bias and protect privacy; robust model risk management to address issues like data quality and hallucinations; and human oversight to ensure final decisions remain with experts, not algorithms.

The IMF recommends that central banks integrate AI into stress testing and macroprudential oversight to mitigate these risks. It also highlights the role of central bank digital currencies (CBDCs), where AI optimizes transaction security and fraud detection, with China’s digital yuan as a case study (IMF, 2025). Yet, unequal AI access across countries—only 20% of low-income nations have AI-ready infrastructure—challenges global financial coordination. The IMF proposes capacity-building initiatives and knowledge sharing to bridge this gap. For central banks, adopting AI requires balancing innovation with ethical considerations, such as addressing biases in credit scoring algorithms that could exacerbate inequality. It is therefore significant for the central banks to evolve their mandates to address AI’s macroeconomic and social implications, ensuring inclusive growth and stability.

Advanced economies (AEs), with 60% job exposure, leverage robust digital ecosystems, while emerging markets (EMs) and low-income countries (LICs), at 42% and 26% respectively, face barriers like weak infrastructure (IMF, 2025). Only 20% of LICs have AI-ready systems, limiting productivity gains (IMF, 2025). For central banks, this poses challenges: AI-driven productivity in AEs may spur inflation, necessitating tighter monetary policies, while LICs’ lag could destabilize trade balances (BIS, 2025). A 2024 AI-driven market crash underscores risks of algorithmic volatility.

Conclusion

AI and ML are redefining central banking, enabling faster insights, stronger risk management, improved policy communication, and efficient operations. From the Bank of England’s generative AI pilots and RBA’s CBDC experiments to MAS’s Project MindForge and BIS’s Project Aurora, central banks are increasingly leveraging AI to modernize financial systems. Yet adoption remains measured and responsible. Regulators worldwide are united by a common vision: AI must enhance, not replace, human expertise in shaping monetary policy and safeguarding financial stability.

References

Bank for International Settlements. (2025). Market intelligence at central banks. BIS. https://www.bis.org/publ/mc_250922.pdf

Basel Committee on Banking Supervision. (2025). Governance of AI adoption in central banks. BIS. https://www.bis.org/publ/othp90.pdf

International Monetary Fund. (2025). The global impact of AI: Mind the gap. IMF Working Paper. https://www.imf.org/-/media/Files/Publications/WP/2025/English/wpiea2025076-print-pdf.ashx