The Bermuda Monetary Authority has launched a consultation on guidance explaining how existing regulatory and supervisory requirements apply when regulated financial services entities develop, procure or use artificial intelligence. The principles-based, technology-neutral framework would not create new obligations or a separate AI approval regime. Instead, entities would apply existing governance, risk management, conduct, cybersecurity, operational resilience, outsourcing, anti-money laundering and recordkeeping frameworks according to each AI use case’s materiality and risk. The proposed guidance covers traditional machine learning, generative AI and agentic AI, including systems supplied by third parties or group entities and AI embedded in other products or processes. Regulated entities would remain accountable for resulting activities and outcomes, with proportionate expectations for lifecycle governance, data controls, validation, meaningful human oversight, explainability, resilience, third-party assurance and supervisory evidence. Higher-impact or autonomous uses would warrant stronger controls, while lower-risk productivity tools could be managed through simpler, platform-level arrangements. Comments are due by Oct. 30, 2026. The authority will consider responses before finalizing the guidance and may develop supplementary examples, supervisory tools or sector-specific materials.

Original source