- Dawgen Global, a Caribbean AI governance advisory firm, states plainly that financial services leads the region in AI deployment and trails it in AI governance. It built its own AEGIS framework because no regulator had one.
- The CTU's Caribbean AI Task Force released an interim report in December 2025 naming financial services a priority sector, with a regional model law still only a recommendation. The Final Report lands at the 2026 Caribbean AI Forum.
- The University of the West Indies launched I-INSIGHT on 8 May 2026, backed by $5 million from Sagicor Financial Corporation, with a Financial Services AI Hub rolling out across all UWI campuses from August 2026.
- ECLAC data puts the region at 6.6% of global GDP but only 1.12% of global AI investment, meaning the capital that does arrive is concentrating on institutions that can prove they govern what they deploy.
- Every Caribbean bank of any size already uses AI. The diligence question that actually separates institutions now is whether governance predates deployment or was retrofitted after the fact.
Caribbean banks are not waiting for a regulator's permission to automate credit decisions. Across Jamaica, Trinidad, Barbados, and the wider CARICOM bloc, financial institutions are running AI models for credit scoring, fraud detection, customer onboarding, and portfolio management in production, today, on live customer data. Dawgen Global, the Jamaica-based professional services firm that advises Caribbean banks, insurers, and credit unions on AI governance, has stated it without hedging across a series of client advisories: no sector in the Caribbean is deploying AI more aggressively, or governing it less rigorously, than financial services. That assessment comes from the firm Caribbean financial institutions themselves hire to fix the problem, not from a critic outside the industry.
The gap matters to anyone with capital in, or capital going into, a Caribbean bank, insurer, or credit union in 2026. Regulatory tightening is coming, agentic AI is already inside loan books, and the region's own coordinating body has confirmed, in writing, that the harmonised rulebook does not exist yet. What follows is where that gap actually sits, who is moving to close it, and what a serious diligence process now has to check.
The Sector Moving Fastest Is Governing It the Least
Financial services has the cleanest data of any Caribbean sector: structured transaction histories, digitised KYC records, claims files, remittance flows. That is precisely why it is also the sector where AI adoption moved past pilots first. A bank does not need to build a data pipeline from scratch to run a credit-scoring model. It already has the data sitting in a core banking system. Tourism, agriculture, and logistics, by comparison, are still working through fragmented, partly paper-based records. Financial services skipped that step.
Skipping the data problem did not mean skipping the governance problem. Dawgen Global's advisory work, and the AEGIS framework it built specifically because no Caribbean regulator had published an equivalent, exists because banks asked for it after they had already deployed, not before. Credit scoring models that touch loan approval decisions, fraud detection systems that flag or clear transactions automatically, and onboarding tools that decide who gets an account and who gets referred for manual review, are running in Caribbean financial institutions right now without a regional standard telling any of them what "governed" is supposed to look like.
What the Region's Own Task Force Actually Recommended
The Caribbean Telecommunications Union's mandate for a regional AI Task Force was approved at its 31st General Conference in October 2025. The Task Force, more than 35 experts across government, academia, the private sector, and civil society, released an interim report on 13 December 2025 that names financial services as one of five priority sectors for targeted AI development support, alongside agriculture, disaster risk reduction, tourism, and health.
The report's five recommended action areas are a regional AI policy framework with model laws and shared oversight, a proposed Caribbean Data Commons for stronger data governance, targeted support in the priority sectors, investment in human capacity to close the region's AI skills gap, and continued consultation through the Caribbean AI Forum. Read that list carefully and the sequencing is the point: the model law is a recommendation for future work, not a standard that exists. The Final Report is due at the 2026 Caribbean AI Forum. Until it lands, and until CARICOM member states individually legislate whatever it proposes, Caribbean financial institutions are operating in the same regulatory space the region's own market thesis describes as technology-positive: open for product development, and correspondingly open on enforcement.
Individual central banks are not silent on this. The Bank of Jamaica and the Central Bank of Trinidad and Tobago both run regulatory sandbox programmes and have signalled openness to AI-augmented financial services, provided institutions can show sound governance. But a sandbox is a conversation with one regulator about one product. It is not a rulebook that travels across borders, and it is not what an acquirer evaluating a multi-territory financial group can point to as evidence of a governed AI programme.
"A bank that can produce its AI governance framework on request is competing for capital against a bank that cannot. In a region receiving barely a tenth of its proportional share of global AI investment, that difference decides which institution gets funded first."
Dr S Budall, Contributing Analyst, Maestro AI Labs
The Capital Is Already Moving. Governance Is Catching Up.
The clearest sign that Caribbean financial services is treating this gap as real, rather than theoretical, is where new capital is landing. On 8 May 2026, the University of the West Indies launched the Institute for Intelligent Systems Governance and Human-Centered Technology, known as I-INSIGHT, backed by a $5 million investment from Sagicor Financial Corporation, one of the region's largest financial services groups. The institute's first operational unit, the Sagicor UWI AI and Financial Services Hub, begins rolling out across all UWI campuses in August 2026. According to UWI leadership, its purpose is to help regional governments build the "regulatory architecture" that foreign AI vendors were never built to provide, because those vendors were not designed around Caribbean credit files, Caribbean informal savings structures, or Caribbean regulatory relationships.
That a major Caribbean insurer and financial group is now the funder of a governance research institute, rather than waiting for a regulator to require one, says something specific about where the sector believes the risk actually sits. Cloud Carib, the Bahamas-headquartered cloud provider, is moving capital into the same region in parallel: over $7 million into its Caribbean expansion through 2025, an AI lab opened in the Bahamas, and new Sovereign Data Center Pods commissioned in Bermuda, Curaçao, and Guyana. Infrastructure and governance are being funded in parallel, by different actors, on different timelines, and a bank's AI programme sits exposed in the gap between the two until both catch up to where deployment already is.
Against that backdrop, the ECLAC figures on regional AI investment are not a footnote. The Economic Commission for Latin America and the Caribbean has documented that the region holds roughly 6.6% of global GDP but attracts only about 1.12% of global AI investment. Capital that scarce does not spread evenly; it concentrates on institutions that can document what they have built. An unrecorded, unaudited AI credit model is therefore a capital-allocation disadvantage as much as a compliance exposure, measured against the institution down the road that kept a paper trail.
What Serious Diligence Now Has to Ask
An investor, acquirer, or reinsurer evaluating a Caribbean financial institution's AI exposure in 2026 needs answers to a specific set of questions, not a general assurance that "AI governance is a priority." The table below is what that conversation should actually cover.
| Diligence Question | Why It Matters | Red Flag |
|---|---|---|
| Who signs off on a new AI model before it touches a customer decision? | Determines accountability if a model discriminates or fails | No named individual or committee; the answer is "the vendor" |
| Does bias testing happen before deployment or only after a complaint? | Before-the-fact testing is governance; after-the-fact testing is damage control | Testing only triggered by a regulator's or ombudsman's inquiry |
| Can the institution produce an audit trail for one specific automated decision on request? | This is the test a regulator or litigant will actually run | Logs exist but cannot be reconstructed into a decision narrative |
| Does the governance framework predate the AI system, or was it written after deployment? | Retrofitted governance is documentation, not control | Framework dated after the earliest production model went live |
| Does the institution map to a named framework (AEGIS or an equivalent) rather than an internal memo? | A named, external framework is testable; an internal memo is not | No framework reference; governance described only in prose |
None of this requires waiting for the CTU's Final Report or a CARICOM model law to start applying. The private-sector frameworks already exist, Dawgen Global's AEGIS model chief among them, and the institutions adopting one now are the ones that will face the smallest compliance shock whenever the regional standard finally arrives. The Caribbean AI Risk Management Council tracks exactly this kind of institutional readiness across the region, and the Caribbean AI Association coordinates the industry conversation feeding into the CTU's process. Neither substitutes for a regulator, but both are functioning as the connective tissue while the formal architecture is still being drafted.
Where the Opportunity Actually Sits
The governance gap is a risk for anyone underwriting an unprepared institution. It is also, distinctly, an opportunity for capital that wants exposure to Caribbean financial AI without inheriting that exposure. Institutions that adopted a named governance framework before regulation forced the issue are the ones best positioned to absorb a CTU model law once it exists, because their controls were built to survive an audit, not to survive a single sandbox conversation. That is the same distinction Maestro AI Labs has documented in board-level AI training data: institutions that governed before they scaled cut vendor spend by more than 70% and stood up governance in roughly half the time of institutions that governed after the fact. The pattern in banking is the same pattern, with higher stakes, because the decisions an ungoverned model makes are credit decisions, not internal productivity gains.
For funds and development finance institutions with a financial inclusion mandate in the region, this changes what underwriting has to establish. Every bank of any size in the Caribbean already uses AI, so what actually needs establishing is whether governance existed before that AI did, or is being written for the diligence file in front of them. The $380 billion Caribbean and LATAM credit gap that Credit Garden and comparable AI credit platforms are built to close only closes responsibly if the institutions extending that credit can show their work. Capital that prices this correctly now will hold the better book when the region's model law finally lands.
