Consultation · BMA · 9 Feb 2026
The Bermuda Monetary Authority summarises feedback on its July 2025 AI discussion paper and confirms it introduces no new AI-specific regulatory requirements at this time. AI governance may be embedded within existing enterprise risk management, governance, conduct, cyber, operational resilience and third-party frameworks; mandatory governance structures would be unsuitable. Proportionality remains anchored in the five risk considerations in paragraphs 60-68 of the discussion paper. The Authority will clarify expectations on proportionality, accountability and assurance, and consider whether sector-specific guidance is warranted for higher-risk use cases.
9 February 2026
Dear Stakeholders:
The Bermuda Monetary Authority (Authority or BMA) would like to thank all stakeholders who provided feedback on the Discussion Paper (DP) entitled The Responsible Use of Artificial Intelligence in Bermuda’s Financial Services Sector, published in July 2025.
The Authority received submissions from a broad range of regulated entities, industry associations and other interested parties across Bermuda’s financial services sector. The feedback was constructive, detailed and reflective of the diverse business models and risk profiles operating within Bermuda’s predominantly international and institutional market.
This letter summarises the principal themes arising from stakeholder feedback and outlines how these themes are informing the Authority’s ongoing policy development. It is not intended to introduce new regulatory requirements specific to Artificial Intelligence (AI) at this time. The BMA’s approach remains outcomes-focused and will provide stakeholders with the requested guidance on managing AI-related risks within the existing regulatory framework, using the principles outlined in the DP.
Stakeholders expressed broad support for the Authority’s proposed outcomes-based and principles-led approach to AI governance. Respondents generally agreed that AI can deliver meaningful benefits when deployed responsibly, including improvements in efficiency, risk management, compliance and market integrity. Many stakeholders also cautioned against overly prescriptive or technology-specific requirements, noting the pace of technological change and the risk of rigid rules becoming outdated quickly.
The BMA notes this feedback and confirms its intention to maintain a fit-for-purpose, technologyneutral supervisory approach. In developing its response to stakeholder feedback, the Authority has also considered various relevant regulatory and supervisory publications issued since the DP, to ensure that its overall direction remains, where appropriate, aligned with emerging international approaches relevant to Bermuda’s market context.
A consistent theme was the view that AI-related risks are largely addressed through existing regulatory frameworks, including corporate governance, conduct, risk management, cyber risk, operational resilience and third-party oversight requirements. Stakeholders encouraged the BMA to avoid duplicative AI-specific structures and instead integrate AI considerations within established enterprise risk management frameworks.
The Authority agrees that AI governance may be embedded, where appropriate, within existing organisational structures that meet supervisory expectations, with enhancements focused on areas where AI introduces materially new or heightened risks. The Authority notes that this approach is consistent with the direction taken by several international counterparts, which have emphasised the application and clarification of existing regulatory expectations to AI use, rather than the creation of parallel or standalone AI regulatory regimes.
Stakeholders broadly supported the Authority’s risk-based and proportionate approach to AI governance. Some respondents suggested that proportionality should more explicitly recognise the expected benefits or competitive advantages of AI adoption. In contrast, others emphasised that AI governance should extend well beyond traditional operational risk considerations, given the potentially transformative impact of AI-enabled business change.
The BMA acknowledges that boards and senior management consider strategic benefits when determining whether to pursue AI-enabled solutions and that AI may amplify or accelerate existing risk drivers across operational, conduct, data, cyber and strategic risk categories. However, from a supervisory perspective, responsible AI governance is anchored in risk and impact rather than commercial upside and does not necessarily require creating governance frameworks that sit outside of or above established enterprise risk management arrangements. Proportionality is intended to align the nature and intensity of governance, controls and assurance with the risk profile and impact of an AI use case. Therefore, mandatory governance structures would be unsuitable. The ultimate accountability for AI outcomes remains unchanged, irrespective of the perceived value or innovation potential of a particular application.
This clarification is consistent with the five key risk considerations set out in paragraphs 60–68 of the DP, which remain the primary basis for applying proportionality across AI use cases.
The Authority will aim to clarify expectations regarding the application of proportionality as part of the next stage of review and industry engagement.
Stakeholders highlighted the importance of reflecting Bermuda’s predominantly wholesale, institutional and cross-border market activity. Respondents cautioned against directly adopting retail-focused AI regulatory approaches developed in other jurisdictions without appropriate adaptation.
The Authority acknowledges this feedback and will continue to differentiate expectations where appropriate, particularly in relation to transparency, explainability and disclosure, while maintaining consistent core governance principles across sectors.
Stakeholders generally supported the Authority’s emphasis on board accountability for AI outcomes. Submissions noted that boards typically exercise oversight through established governance arrangements, supported by management committees, risk functions and internal audit. Feedback also highlighted the importance of proportionate AI literacy at the board and senior management level.
The BMA will aim to clarify expectations around accountability, delegation and assurance mechanisms, recognising that while boards are not expected to perform technical reviews, they must maintain sufficient AI literacy to discharge their duties. Examples include effectively challenging management, setting risk appetite and overseeing AI risks, all with the depth of understanding expected to be proportionate to the materiality and risk of AI use.
Some stakeholders noted the growing use of AI within investment, trading, market surveillance and research functions, and the potential implications for market integrity and financial stability. The feedback highlighted risks associated with correlated or “herding” behaviours, accelerated market dynamics under stress, reliance on alternative or unstructured data sources, and the potential for AI-enabled market manipulation or misleading disclosures regarding AI capabilities.
The Authority also notes recent international supervisory commentary issued after the publication of the DP, that highlights the potential for unintended or emergent behaviours arising from interactions between independently deployed AI systems, particularly in capital markets contexts.
These considerations will inform the BMA’s ongoing assessment of whether sector-specific guidance or supervisory clarification is warranted for higher-risk use cases, which may build on existing market integrity, conduct and any other relevant regulatory frameworks.
Stakeholders strongly emphasised the importance of international regulatory alignment, particularly for firms operating under group-wide AI governance frameworks that must comply with multiple regulatory regimes.
The BMA notes the operational challenges associated with fragmented or inconsistent international AI requirements. The Authority will continue to monitor global developments and engage with international partners and peer regulators to support coherence where practicable.
Stakeholders acknowledged that many AI capabilities are delivered through third-party providers and highlighted challenges relating to transparency, contractual access, auditability and vendor concentration. Respondents generally considered that existing outsourcing and third-party risk management frameworks provide a strong foundation for managing these risks.
In this context, the Authority will consider whether further supervisory focus is required to identify and map dependencies on critical AI-related service providers, including cloud; model and data infrastructure providers, as part of existing third-party and operational resilience expectations.
Respondents identified practical challenges related to skills, resourcing, independent validation and integration with existing systems, particularly for smaller or less complex institutions.
The Authority notes the importance of phased implementation, supervisory engagement and continued dialogue to support the proportionate and effective adoption of AI governance practices. The BMA will also continue to consider its own supervisory tools, capabilities and international cooperation arrangements, recognising that effective oversight of AI-enabled activities will evolve alongside market adoption and supervisory practice globally.
As set out in the DP, the Authority’s current focus is on monitoring how existing governance, risk management, outsourcing, market conduct, cybersecurity and operational resilience frameworks are applied to AI use across the financial services sector.
The BMA will continue to engage with stakeholders and monitor international developments to inform its analysis of existing regulatory requirements, current market practices, and the principles outlined in the DP to ensure the frameworks remain fit-for-purpose.
Where this analysis signals the need for enhancements to the existing regulatory framework, they would be expected to be developed incrementally and proportionately, and subject to the Authority’s normal consultation process. The Authority’s objective is to support responsible innovation while ensuring that any enhancements to supervisory expectations are practical, risk-based and avoid unnecessary duplication or unintended regulatory burden.
The Authority thanks stakeholders for their engagement and constructive feedback. Consultation between the BMA and stakeholders will be essential in developing a supervisory approach to AI that supports innovation while safeguarding financial stability, market integrity and appropriate customer outcomes.
Sincerely,
Bermuda Monetary Authority
Note from RegAlert. AI assistants can read this document in full, and search 70,000+ more, through the RegAlert MCP connector (https://mcp.regalert.today/mcp). Free with an account. How to connect ChatGPT, Claude or Cursor.
More like this from BMA
BMA published 4 documents in the last 30 days. We email you each new one the day it's published.
Source: Bermuda Monetary Authority — original document · Summary generated with machine assistance and checked by an AI reviewer before publication; the authoritative text is the regulator's original document. How RegAlert works
In RegAlert: added