IMF to central banks: Tighten AI oversight to safeguard financial stability

…Warns unchecked AI could amplify market shocks, cyber risks

By Uche Usim

The International Monetary Fund (IMF) has urged central banks and financial regulators worldwide to strengthen oversight of artificial intelligence (AI) as its growing adoption across the financial system creates new risks that could threaten global financial stability.

In a new IMF Blog authored by the Fund’s Financial Counsellor and Director of the Monetary and Capital Markets Department, Tobias Adrian, the IMF said AI is rapidly transforming how financial institutions price risk, allocate credit, monitor markets and respond to financial shocks.

While the technology is improving efficiency, lowering costs and enhancing decision-making, the IMF warned that inadequate governance could allow AI-driven activities to amplify market volatility, cyber threats and systemic risks.

According to Adrian, AI has become deeply embedded in the decision-making architecture of modern finance, making it increasingly important for central banks and supervisors to ensure that its deployment strengthens, rather than weakens, financial stability.

He identified three immediate priorities for policymakers: strengthening oversight of AI-driven trading, lending and supervisory technology; improving visibility into AI use and concentration risks; and expanding international cooperation on cyber resilience and operational stability.

The IMF noted that AI is compressing time and distance in financial markets as trading, lending decisions and regulatory analysis increasingly occur in real time. While this enhances market efficiency during normal conditions, it could also accelerate the spread of financial shocks during periods of stress.

According to the Fund, AI-powered trading systems are now widely used by investment banks, hedge funds and asset managers to analyse earnings reports, regulatory filings and economic data within seconds.

Although AI has so far improved market liquidity, reduced transaction costs and accelerated price discovery, the IMF warned that the widespread use of similar machine-learning models could intensify market swings if many firms react simultaneously to the same signals.

The Fund’s analysis showed that some AI-driven investment funds rebalance their portfolios much faster than conventional funds, increasing the possibility of synchronized trading that could worsen market volatility.

It cautioned that future flash crashes may no longer result mainly from programming errors but from numerous AI systems responding simultaneously to identical market information.

The IMF also highlighted the challenge of transparency, noting that even sophisticated financial institutions often struggle to explain how AI models make decisions during periods of market stress.

To address these concerns, the Fund said regulators should strengthen monitoring of AI-based trading strategies, improve data collection on AI adoption and model dependencies, and incorporate AI-related scenarios into financial stress tests.

Beyond financial markets, the IMF said AI is transforming the operational backbone of banks, payment systems, stock exchanges and clearing houses by automating risk management and system monitoring.

However, it warned that the growing dependence on a small number of cloud computing, data and AI model providers creates significant concentration risks.

According to the IMF, a disruption affecting one major technology provider—whether caused by technical failures, cyberattacks or geopolitical developments—could simultaneously affect numerous financial institutions.

The Fund noted that authorities such as the European Central Bank and the Bank of England have already expanded operational resilience frameworks to include AI and cloud service providers, urging other jurisdictions to adopt similar safeguards.

The IMF further observed that AI is reshaping financial supervision itself through Supervisory Technology (SupTech), enabling regulators to analyse large volumes of market data, detect anomalies and identify emerging risks more efficiently.

Several central banks, including those of France, Germany, Portugal and Japan, already deploy machine-learning tools in market surveillance, while the US Federal Reserve, European Central Bank and Bank of Canada use natural-language processing to analyse supervisory reports and consumer complaints.

Despite these advances, Adrian stressed that AI should complement—not replace—human judgment.

He warned that over-reliance on automated systems without adequate expertise and governance could create new supervisory blind spots, particularly during financial crises.

The IMF also identified AI-enhanced cyber threats as another growing concern.

It said generative AI is making phishing attacks more convincing, enabling fraud schemes to evolve in real time and shortening the time between the discovery and exploitation of software vulnerabilities.

The Fund therefore called on central banks to strengthen cyber resilience, conduct AI-focused stress exercises and invest in defensive AI technologies capable of countering increasingly sophisticated attacks.

According to Adrian, AI has become a financial stability issue that cuts across markets, financial institutions, infrastructure and regulation.

He said early policy action, stronger international collaboration and robust governance frameworks would enable countries to harness AI’s benefits while reducing systemic risks.

“If policymakers act early and collectively, AI can reinforce global financial resilience. If they do not, future instability may be faster, more correlated, and harder to manage than past episodes,” Adrian said.