Central Banks Using AI For Low-Risk Processes, Wary Of Cybersecurity
Around 6% of central banks surveyed say AI is not yet supporting their operations in any meaningful way.
A small survey of central banks across the world has revealed most are experimenting with artificial intelligence only in low-risk processes, even as they remain cautious about cybersecurity risks.
The UK-based independent think tank OMFIF said central banks are approaching AI with caution, shaped by years of market shocks, cybersecurity incidents and the hard lessons of operational risk.
A working group, made up of BNY, Bridgewater and Capital Group, brought together 10 central banks through bilateral exchanges across Europe, Africa, Asia and Latin America. The think-tank did not disclose the names of the participating central banks.
Most institutions are experimenting with AI only in low-risk processes like scanning market news, flagging anomalies and summarising reports.
According to OMFIF’s Global Public Investor 2025 survey, 61% of central banks say AI is not yet supporting their operations in any meaningful way. They describe AI as a practical convenience rather than a strategic tool and acknowledge that adopting AI without airtight governance could endanger operations and credibility.
"Model reliability remains a top worry, especially given how often AI tools misinterpret unusual scenarios. Central banks, whose work revolves around rare but disruptive shocks, have little tolerance for such risks," the report said.
The survey also found some institutions have in-house data scientists and secure enterprise environments. Others have small teams, limited budgets and governance structures that make experimentation slow.
"Central banks have no interest in outsourcing judgement to machines. The working group report makes clear that human oversight is the anchor. AI can summarise, filter and accelerate, but decisions remain with people," the OMFIF report said.
Artificial intelligence has disrupted most modern industries, including finance. Rapid development has provided tools that mimic human intelligence to automate processes and enhance decision-making. Key applications include enhanced risk management, fraud detection, algorithmic trading, and operational efficiency.
The proliferation of AI has come even as regulation lags in most countries, raising concerns over privacy, data security and impact on jobs.
