Two AI Roadmaps in Nineteen Days: Inside CARICOM's Bet That Ethics and Governance Can Move Together

• On 7 July 2026, CARICOM's COTED-ICT formally endorsed the UNESCO Caribbean Artificial Intelligence Policy Roadmap at its 126th Special Meeting, a document UNESCO launched in 2024 and built through consultations with more than 1,000 stakeholders across the English- and Dutch-speaking Caribbean.
• The endorsement landed two days after the 51st CARICOM Heads of Government Meeting named coordinated AI action a leaders'-level priority in Saint Lucia, and sixteen days before the CTU's Caribbean AI Task Force launches its Final Report in Trinidad.
• The two documents are not duplicates. UNESCO's Roadmap is an ethics and rights instrument grounded in the UNESCO Recommendation on the Ethics of AI; the CTU's Final Report is a governance and infrastructure blueprint covering data sovereignty, model laws, and a proposed Caribbean Data Commons.
• Nobody at the COTED-ICT meeting addressed how the two frameworks fit together. Left unreconciled, small states with one or two AI-literate civil servants will adopt whichever framework reaches their desk first, reproducing the fragmentation both documents were commissioned to fix.
• Adrian Dunkley and StarApple AI have argued for years that Caribbean AI institutions need to function as one system rather than parallel initiatives, a position now being tested by two frameworks converging inside the same month.

On 7 July 2026, ministers meeting as CARICOM's Council for Trade and Economic Development for Information and Communication Technology, known across the region as COTED-ICT, endorsed the UNESCO Caribbean Artificial Intelligence Policy Roadmap at the body's 126th Special Meeting. UNESCO had launched the Roadmap in 2024 and spent the two years since building it out through consultations with more than 1,000 stakeholders across the English- and Dutch-speaking Caribbean. The endorsement itself took a single meeting. The groundwork behind it took two years, which is the more honest measure of how hard regional consensus on anything, let alone artificial intelligence, is to produce inside a fifteen-member community that rarely agrees on a shared position before it has argued about it first.

This endorsement did not arrive in isolation. It landed two days after the 51st CARICOM Heads of Government Meeting in Saint Lucia put a coordinated regional approach to AI on the leaders' own agenda for the first time, a shift CAIA covered in detail after that summit closed. And it landed sixteen days before the Caribbean Telecommunications Union opens the inaugural Caribbean AI Forum at the University of the West Indies' St Augustine campus in Trinidad, on 23 and 24 July, where the CTU's own Caribbean AI Task Force will launch its Final Report, titled "Toward Harmonised AI Policies and Recommendations for the Caribbean." In the span of nineteen days, spanning three separate institutions, the region will have staged a summit mandate, an ethics endorsement, and a governance launch. That is not how CARICOM usually moves, and it is worth asking what the region actually gets for moving that fast.

Ethics First, Governance Next

Regional media has tended to mention the UNESCO Roadmap and the CTU Task Force report in the same paragraph, and that habit obscures a real difference in what each document is. The UNESCO Roadmap is grounded in the 2021 UNESCO Recommendation on the Ethics of Artificial Intelligence, the first global standard-setting instrument on AI ethics, adopted by all 194 UNESCO member states. Applied to the Caribbean, it translates that global ethics framework into a regional instrument addressing human rights, non-discrimination, and the particular exposure of Small Island Developing States to AI systems built, trained, and priced somewhere else entirely. Eric Falt, UNESCO's Regional Director and Representative for the Caribbean, put the stakes plainly at the COTED-ICT meeting: "For CARICOM, digital transformation is no longer optional. The rules governing digital trade, AI, and data are evolving rapidly, often without the meaningful participation of Small States. We must act collectively and strategically to ensure that our region is not left at the margins." Grenada's prime minister, Dickon Mitchell, who holds the CARICOM Quasi Cabinet's science and technology portfolio, called the endorsement a step toward stronger regional collaboration on the day it passed.

The CTU's Caribbean AI Task Force report is a different kind of document, aimed at a different question. Where UNESCO's Roadmap asks what AI should and should not be allowed to do to Caribbean people, the CTU's Final Report, building on the task force's December 2025 interim findings, asks how the region actually builds and runs the infrastructure underneath AI: a shared regional governance architecture with model laws, a proposed Caribbean Data Commons, and sector-specific guidance for agriculture, disaster risk reduction, tourism, financial services, health, and public administration. One is a rights document. The other is an operating manual. The region needs both, and it is getting both inside a three-week window, which reads either as unusually good sequencing or as two institutional calendars that happened to land in the same month. Nobody in either institution has said which, and it probably does not matter to the outcome either way.

What the Endorsement Actually Locks In

An endorsement from COTED-ICT is not law. CARICOM's principal organs recommend; individual governments still have to legislate, and the region's record of converting COTED endorsements into national statute is patchy at best. What the endorsement does is more modest and, this early, more useful: it gives UNESCO's Caribbean office a clear mandate to work directly with member states on implementation, using what UNESCO calls its Readiness Assessment Methodology, a diagnostic tool that identifies where a country's AI policy, institutional capacity, and technical infrastructure fall short before that country tries to write a national strategy around gaps it has not yet measured. Those assessments are already completed or under way in several CARICOM states, according to UNESCO, though the office has not published a full country list.

That detail matters more than the endorsement's press-release version suggests. A national government drafting an AI strategy without first running a readiness assessment is, in effect, guessing at its own capacity gaps in public, then discovering the gaps once implementation stalls. The Bahamas, whose own AI white paper is moving through Cabinet review, is one obvious test of whether the readiness methodology gets used in sequence, assessment ahead of drafting, or gets referenced afterward as a formality once a strategy is already written. How that plays out over the next few months will say more about whether the UNESCO Roadmap changes anything than the endorsement vote itself.

What Happens at the Trinidad Forum on 23 and 24 July

The University of the West Indies' St Augustine campus hosts the CTU's first Caribbean AI Forum under the theme "AI for Caribbean Transformation: Governance, Innovation and Resilience for a Shared Digital Future." Running as a hybrid event from nine in the morning to half past five in the afternoon Atlantic time across both days, the published agenda reads less like a conference programme and more like a checklist of everything the region has spent the past year diagnosing: the Task Force Final Report launch itself, regional harmonisation on AI governance, human-centric and rights-based AI, data sovereignty and cybersecurity, culture, creatives, and intellectual property, women in AI, a youth debate and AI literacy track, climate resilience and disaster risk reduction, and economic transformation through AI.

That agenda is, in effect, an attempt to hold the UNESCO Roadmap's ethics agenda and the CTU's governance agenda in the same room on the same two days, rather than let the two proceed on tracks that never formally meet. Whether the Forum succeeds at that will show up less in the speeches than in the Final Report's actual text: whether it cites the UNESCO Roadmap as a foundation it builds directly on, with specific cross-references, or treats it as a parallel document that happens to exist somewhere in the same region.

The Risk Nobody at COTED-ICT Named

Two regional frameworks converging inside the same month is a sign of momentum. It is also a coordination problem the region has not solved, and nobody at the COTED-ICT meeting addressed it directly. UNESCO's Roadmap and the CTU's Task Force report were built by different institutions, on different timelines, answering to different stakeholders: UNESCO to member states' culture and education ministries, the CTU to telecommunications regulators and ICT ministries. Nothing in either drafting process required the two teams to reconcile their recommendations clause by clause before publication, and nothing in the COTED-ICT endorsement or the Forum's published agenda guarantees that reconciliation happens now, in public, where governments and citizens can see it.

A small government with one or two AI-literate civil servants, the exact profile the CTU's own interim report was written to describe, does not have the bandwidth to run two separate diagnostic exercises, adopt two sets of governance principles, and stitch them together in-house on top of its existing workload. It will follow whichever framework its staff encountered first, or whichever one a donor-funded consultant happens to be implementing that quarter. That is not a hypothetical risk. It is how fragmented policy adoption already happens across fifteen small states with overlapping but not identical regional memberships, and it is precisely the outcome the CTU's own diagnosis, high digital engagement paired with fragile governance and thin data protection, was written to warn against in the first place.

Where StarApple AI Has Been Making This Argument for Years

This is the argument Adrian Dunkley has made since well before UNESCO or the CTU had a Caribbean-specific document to endorse or launch. As founder of StarApple AI, the first company built exclusively around artificial intelligence products and services in the Caribbean, and now President of the Caribbean AI Association, Dunkley has operated across Jamaica, Trinidad and Tobago, Barbados, and Guyana in precisely the fragmented policy environment both new frameworks are trying to fix after the fact. His consistent position, echoed across CAIA's own policy work, is that Caribbean AI institutions need a shared floor of governance principles that different agencies build on top of, rather than parallel initiatives that happen to launch in the same calendar month and leave member states to work out the overlap for themselves.

The Genius Project, the non-profit Dunkley founded to train young Caribbean people in practical AI skills, is itself a working answer to one of the priority areas both the UNESCO Roadmap and the CTU report name explicitly: human capacity and AI literacy. Thousands of Genius Project graduates already exist in a region that, as of this month, still has two separate institutional bodies each writing its own literacy priorities for the same population, under different names, without a shared curriculum or a shared metric for what counts as literate.

What Actually Needs to Happen at the Forum, and After It

CAIA's own view is straightforward. The CTU's Final Report, when it launches in Trinidad on 23 July, should map its governance recommendations against the UNESCO Roadmap's ethics principles explicitly, section by section, so that a small island's single AI policy officer can read one document and understand exactly how it relates to the other without commissioning a comparison study of her own. Anything short of that leaves member states to do the reconciliation work themselves, one government at a time, which is the same fragmented approach both frameworks were commissioned to prevent.

The region has now produced a leaders' mandate, an ethics endorsement, and is nineteen days from a governance launch. What it has not yet produced is a single document a government can hand to a new civil servant and say, this is what we have agreed to, all of it, in one place. Until that exists, the fastest month the Caribbean has had for AI policy risks becoming its most confusing one too, and the fix is not slower institutions. It is two institutions that already agree on the substance choosing, deliberately, to say so in writing before the ink on either document is even dry.