Central banks must secure trust as AI risks reshape finance: IMF tech chief

International Monetary Fund (IMF) Chief Information Officer Shirin Hamid warned that central banks must embrace artificial intelligence (AI) with strict governance and human oversight to maintain public trust in financial systems amidst emerging technology risks. Speaking at the 76th Anniversary Oration of the Central Bank of Sri Lanka (CBSL) in Colombo, Hamid highlighted AI's potential to facilitate voice cloning, deepfake fraud, and cyberattacks executed at machine speed.

She stressed that a trustworthy AI system in a central bank is one whose AI is explainable, governed, secure, ethical, and subject to human supervision.

Three major AI threats facing financial institutions were identified by Hamid: cloned audio and video impersonating humans, autonomous AI swarms executing attacks at machine speed, and indirect breaches manipulating internal AI assistants within bank firewalls. She revealed that even a three-second audio clip is sufficient for AI to generate an indistinguishable clone.

Hamid also cautioned about data risks, noting that when IMF tests commercial generative AI on economic data, tools such as ChatGPT often produce inaccurate GDP figures up to two-thirds of the time. To address this, the IMF developed StatGPT, a system that queries verified data directly to ensure accurate results.

Hamid outlined four pillars that should ground technological transitions in financial systems: dependable data, trustworthy systems, operational resilience, and adaptive governance. She emphasized that institutions must not view AI solely as an IT department project. According to Hamid, human judgment remains the most critical resource, and for every AI-influenced decision, there should be a named human who is answerable.

She urged that AI should augment judgment, not replace it, and autonomy should only be granted after a track record of trust has been established. Hamid concluded that while financial tools will continue to evolve, the fundamental mission of an intelligent central bank remains unchanged: protecting public confidence from the vault to the algorithm.