FSB Pushes Banks to Build AI Risk Guardrails

The Financial Stability Board is urging financial institutions to set up safeguards against artificial-intelligence-related risks, according to a statement from the group.

The FSB, which coordinates financial regulation for the Group of 20 economies, did not detail specific rules. It said firms should ensure that AI models used in lending, trading, and risk management are explainable, tested for bias, and backed by human oversight. The call comes as banks and hedge funds deepen their use of machine-learning tools for credit scoring, portfolio optimization, and fraud detection.

Regulators globally have warned that a handful of dominant AI vendors could create systemic concentration risk. An error or manipulation in a widely used model could cascade across institutions that rely on the same technology. The FSB's push mirrors earlier guidance from the Basel Committee on Banking Supervision, which flagged model opacity and data dependency as emerging vulnerabilities.

The statement marks the first time the FSB has issued a collective stance on AI governance since the technology's rapid adoption in finance over the past two years. The board said it will monitor implementation and may issue more prescriptive standards if voluntary guardrails prove insufficient.

For traders and risk managers, the practical effect is likely to be incremental. Institutions that already run model validation and explainability frameworks will face less disruption than those deploying AI without documented controls. The FSB's timeline for feedback and follow-up rules was not specified.