- The Caribbean Telecommunications Union hosts the 1st Caribbean AI Forum on 23-24 July 2026 at UWI St. Augustine, Trinidad, to launch the CAITF Final Report, the region's first coordinated AI policy framework.
- The task force behind it, more than 35 experts chaired by Dr Craig Ramlal of UWI's AI Innovation Centre, published an interim report in December 2025 naming five priority actions, including a shared Caribbean Data Commons.
- Independent country tiering shows only Jamaica, Barbados, and The Bahamas with active data protection law and a functional regulator. Six states, Belize, Suriname, Saint Kitts and Nevis, Montserrat, Dominica, and Haiti, have minimal or no AI governance infrastructure at all.
- As of April 2026, zero CARICOM states had published a national AI strategy of their own to sit under the regional framework.
- The gap is not confined to government. A March 2026 PwC survey found just 6% of Caribbean board directors believe their boards spend enough time on AI oversight.
- A framework without local enforcement capacity in six countries is a coordination document first and a governance system second. Maestro AI Labs builds the data and model infrastructure the framework describes, rather than waiting on the policy layer to catch up.
A Forum Built to Close a Two-Year Gap
Governance summits have not been rare this year. Santo Domingo produced a Ministerial Declaration in June. Bangkok produced a sovereignty pledge signed by more than 100 countries in February. Both were genuine diplomatic achievements and neither one told a bank in Kingston or an insurer in Bridgetown what to actually do on Monday morning. The Caribbean AI Forum, convening at UWI St. Augustine in Trinidad on 23 and 24 July 2026 under the banner "AI for Caribbean Transformation," is different in one specific way: it exists to launch a document written for this region, by people who work in it, with a named author list and a named chair rather than a joint declaration assembled by consensus across dozens of foreign ministries.
That document is the Final Report of the CTU Caribbean AI Task Force, known as CAITF. The task force formed in 2025 under a ministerial mandate from CTU member governments and has been meeting since, chaired by Dr Craig Ramlal, who runs the University of the West Indies' Artificial Intelligence Innovation Centre in St. Augustine. Its interim findings, released as "Toward Harmonised AI Policies and Recommendations for the Caribbean" on 13 December 2025, were presented by Kirk Sookram, Deputy CEO of Trinidad and Tobago's Telecommunications Authority. The Final Report due at the forum is meant to convert those findings into something governments can actually adopt.
What the Task Force Is Actually Proposing
The interim report names five priority areas, and they are worth reading in full because each one describes a gap that already has a cost attached to it. The first is a regional AI policy framework: model laws and shared oversight mechanisms that individual legislatures could adapt rather than draft from nothing. The second is stronger data governance and infrastructure, including a proposed Caribbean Data Commons, a shared pool of data governance capacity that small states could draw on jointly instead of each trying to fund a national AI regulator alone. The third is direct support for AI adoption in agriculture, disaster risk reduction, tourism, financial services, health, and public administration, the sectors where the region's own economy actually runs. The fourth is human capacity and AI literacy, closing a skills gap the report treats as a precondition for everything else on the list. The fifth is institutionalising the Caribbean AI Forum itself as a recurring venue, rather than a one-off event.
The report frames the underlying problem as a "connectivity paradox": the region shows strong digital engagement on the surface, while running on fragile infrastructure, high connectivity costs, fragmented regulation, and limited data governance underneath. Most intra-Caribbean internet traffic still routes through infrastructure outside the region, which means a message sent from Kingston to Port of Spain can pass through a foreign jurisdiction before it arrives. Data residency across the region is a patchwork: sensitive government and financial records are mostly kept on-premises, while other sectors, tourism especially, run heavily on global cloud providers with no requirement that the data stay anywhere near the customer it describes.
The Tiering Problem the Report Cannot Paper Over
A regional model law is only as strong as the institution enforcing it locally, and here the numbers get harder to spin. Country-level tiering compiled by regional AI policy analysts sorts CARICOM and wider Caribbean states into three groups by how much functioning AI and data governance infrastructure they actually have today.
| Tier | Countries | What They Have |
|---|---|---|
| Tier 1 | Jamaica, Barbados, The Bahamas | Active data protection law and a functional regulator |
| Tier 2 | Trinidad and Tobago, Saint Lucia, Guyana, Antigua and Barbuda, Grenada, Saint Vincent and the Grenadines | Partial frameworks with operational gaps |
| Tier 3 | Belize, Suriname, Saint Kitts and Nevis, Montserrat, Dominica, Haiti | Minimal or no AI governance infrastructure |
Six states with minimal or no governance infrastructure is not a footnote to a regional framework, it is close to a third of the countries the framework is meant to cover. Dominica has no data protection law on the books at all. Haiti has provisions in law with no institutional capacity behind them to enforce what is written. A model law travelling to either country does not meet a weak regulator, it meets no regulator. This is the same asymmetry that let Huawei Cloud's Model-as-a-Service platform reach production in Jamaica and Trinidad and Tobago in May, a month before Santo Domingo, with no national AI strategy anywhere in CARICOM to weigh it against. A model law solves the drafting problem. It does not solve the capacity problem, and the capacity problem is the one actually stopping enforcement.
None of this is an argument against the framework. A shared Caribbean Data Commons is a genuinely sensible answer to a real constraint: most CARICOM states are too small to fund a standalone AI regulator on their own tax base, so pooling the function is close to the only realistic path to giving Tier 3 states anything to enforce with. The honest read is narrower than "the region has a plan." It is that the region has a plan for a system that, in a third of its member states, still needs to be built from the ground up before the plan can run on it.
The Boardroom Has the Same Gap
Regulators are not the only ones behind. A PwC survey published in March 2026 asked Caribbean board directors whether their boards spend sufficient time on AI oversight. Only 6% said yes. Just 9% said their board receives adequate information to address AI risk. That is a private sector governance gap running in parallel to the public sector one this forum is trying to close, and it matters more to an investor than it might first appear. Regulators eventually catch up to a market; boards that cannot see their own AI exposure keep signing contracts in the meantime.
Put the two gaps side by side. Latin America and the Caribbean attract 1.12% of global AI investment against 6.6% of global GDP, a mismatch that has held for years now. Global enterprise AI adoption reached 72% in 2025, up from 20% in 2017, according to McKinsey. Caribbean companies are adopting the same technology at a comparable pace to everywhere else, largely through the same handful of foreign cloud and model vendors, while the governance layer that is supposed to make that adoption accountable, both inside the boardroom and outside it in the regulator's office, has not moved at anything close to the same speed.
A model law describes what a government intends to require. A functioning regulator is what actually requires it. Between 23 July and whenever six countries have the second thing, the first thing is largely a statement of intent.
Why a Regional Framework Is Harder Than a National One
It is worth being fair to the task force here, because the difficulty is real and not a failure of drafting. A single country writing its own AI law answers to one legislature, one budget, and one enforcement agency it can staff and fund directly. A regional framework has to work across fifteen-plus jurisdictions with wildly different fiscal capacity, from Trinidad and Tobago's energy-funded state apparatus to Dominica's, which cannot currently fund a data protection office at all. Harmonised model law is the correct instrument for that problem: it lets a small state adopt language it did not have to draft, rather than leaving each one to build a regulatory regime from a blank page.
What harmonised law cannot do on its own is fund the office that enforces it. That is where the Caribbean Data Commons idea becomes the load-bearing piece of the whole proposal rather than one bullet point among five. If it is built as genuinely shared infrastructure, with staff, a technical capability, and a mandate that Tier 3 states can plug into rather than replicate, it is the mechanism that turns a model law from an aspiration into something Dominica or Haiti could actually operate. If it stays a paragraph in a report with no dedicated budget attached, the tiering table above will look much the same in July 2027 as it does today.
Building the Layer the Framework Describes
This is the specific gap StarApple AI, the first AI company established in the Caribbean, has been building toward since Adrian Dunkley founded it in Jamaica in 2023, well before a July forum made the governance gap visible to a wider audience. Maestro AI Labs, its research and product arm, ships the alternative directly. Harmonics runs on regional knowledge graphs instead of a generic model with a login screen bolted on top. Credit Garden scores borrowers using data drawn from the cooperative and SUSU networks Caribbean lenders actually use. Data Archaeology is the underlying archive, more than 2.3 million ground-collected records, that makes both possible without depending on a foreign vendor's training set.
None of that competes with what CAITF is trying to build. A regional regulator and a regionally owned data and model layer solve different halves of the same problem, and the region needs both. What Maestro AI Labs offers in the meantime is an answer to the question a Tier 3 government or a Tier 2 bank has to answer today, before a Caribbean Data Commons has staff and a budget: when you adopt an AI system this quarter, whose data trained it, and who actually controls it. Adrian Dunkley has argued this case since before the current wave of summits, and the wider network he built, from AI Trinidad & Tobago building talent and community in the very country hosting this forum, to AI Jamaica doing the same at home, is the case made in practice rather than in a keynote.
What to Watch After 24 July
The forum itself is not the story. What happens in the months after it is. Watch whether the Caribbean Data Commons gets a named budget and a staffing plan, or stays a line in a report. Watch whether any Tier 3 government commits to a specific enforcement timeline rather than a general endorsement of the framework. Watch whether the boards sitting on the PwC survey's 6% start asking their own AI vendors the questions a functioning regulator would eventually ask for them, because in the gap between now and when that regulator exists, the board is the only oversight some of these systems will get. A rulebook that arrives without an enforcement layer in a third of its member states is not a failure. It is the first step in a longer build, and the honest question for anyone backing this region's AI infrastructure is whether they are funding the step that comes next.