Global Regulators Tighten AI Governance for Financial Firms
Financial firms are deploying generative AI across lending, fraud detection, customer service and compliance, exposing them to risks ranging from biased decisions and data leaks to opaque model outputs. Regulators including the Michigan Department of Insurance and Financial Services, the UK Financial Conduct Authority and the Central Bank of the UAE have made clear that using AI does not displace existing legal duties. Governance failures can therefore become examination findings even where no AI-specific statute applies.
Michigan’s regulator on Jan. 14, 2026, directed financial service providers to maintain written AI systems programs covering model inventories, validation, bias testing, continuous monitoring and vendor due diligence. The FCA reiterated on Feb. 13 that it would rely on existing frameworks rather than introduce separate AI rules. On Feb. 23, the CBUAE issued guidance for licensed financial institutions requiring accountability, explainability, meaningful human oversight, data protection and controls over outsourced AI. The measures signal that firms must produce operational evidence of compliance, not policies alone.
All Coverage
1 original reportsThe Backstory
The history behind this eventU.S., UK and EU Regulators Advance AI Risk Management and Live Testing in Financial Services
Banks and fintech companies already use AI for credit underwriting, pricing, fraud detection and anti-money laundering. But model bias, drift, black-box decision-making and concentration among third-party vendors could harm consumers and amplify financial-stability risks. The U.S. Treasury, the UK's Financial Conduct Authority and the European Union are therefore moving beyond broad principles toward audit-ready controls, regulatory sandboxes and live testing.
On February 19, 2026, the U.S. Treasury released an AI risk management framework for financial services and an AI lexicon. Developed with input from more than 100 financial institutions and government agencies, the framework sets out 230 control objectives across governance, mapping, measurement and management. It creates no new legal obligations, and no grant amount was announced. The FCA's second AI Live Testing cohort is scheduled to begin in late April, while the EU classifies credit scoring as a high-risk use under the AI Act.
Subscribe to Mark Radar Weekly
Every Friday, the week's strongest signals in your inbox. Unsubscribe anytime.
If you search news on Google, you can set Mark Radar as a preferred source—our coverage will show up more often in your results. Set as preferred source on Google →