CARICOM Endorsed a Caribbean AI Roadmap. Adoption Is Still Stuck at 13 Percent.

Three days ago, in Georgetown, CARICOM's Council for Trade and Economic Development, sitting in its ICT configuration, endorsed a regional AI policy roadmap. If you did not hear about it, you are not alone. It did not trend. No prime minister posted about it. But it is the most consequential piece of Caribbean AI policy news this year, and I want to be precise about why, because "consequential" and "sufficient" are not the same word.

The roadmap in question is the UNESCO Caribbean Artificial Intelligence Policy Roadmap, drafted in 2024 after consultations with more than 1,000 people across the English and Dutch-speaking Caribbean: civil servants, university researchers, private sector founders, and civil society groups. On July 7, COTED-ICT, the 126th special meeting of that council, gave it a formal regional stamp. For the first time, every CARICOM member state has agreed, on paper, to a shared approach to how AI gets governed, taught, and deployed across the region.

What Actually Happened in Georgetown

Strip away the acronyms and the meeting was simple. UNESCO's Caribbean office presented a document. Ministers and trade officials from CARICOM's fifteen member states reviewed it and signed off. That is the entire mechanical event. No new agency was created. No budget line was appropriated. No country committed to a specific piece of legislation by a specific date.

What did happen is harder to dismiss. Getting fifteen governments, ranging from Trinidad and Tobago's oil-financed civil service to Montserrat's population of under 5,000, to agree on a single AI policy framework is not trivial. CARICOM has a long, frustrating history of regional agreements that individual member states quietly decline to implement. An endorsement at COTED-ICT is the first formal step toward making AI policy a shared regional commitment rather than fifteen separate improvisations, and given how thin most Caribbean governments' technical capacity actually is, a shared starting template saves the smaller territories from building a national AI framework from a blank page with no staff to do it.

I have spent the better part of two decades building AI capacity in this region, first as a physicist, then as the founder of StarApple AI, the Caribbean's first AI company. I did that work before any roadmap existed, because the region could not wait for one. So my read on this endorsement is not cynicism about process. It is a specific, informed judgment about what a document like this can and cannot do on its own.

The Four Pillars, and Which Ones Have Teeth

The roadmap organizes itself around four pillars. Culture and Creativity addresses what AI does to Caribbean music, language, and creative industries, an area that gets almost no attention in most national AI strategies and genuinely should. Governance and Transformation covers regulation, data protection, and how public sector agencies actually use AI in daily operations. Upskilling and Education is about building AI literacy from primary school through the workforce. Resilience and Sustainability ties AI to disaster management and climate adaptation, which for a hurricane belt is not an abstract add-on but close to the whole point. On paper, all four pillars are sound. In practice, they carry very different amounts of weight. Governance and Transformation is the pillar governments can act on unilaterally: a ministry can draft a data protection amendment or issue a public sector AI use policy without waiting on anyone else. Upskilling and Education requires sustained budget over years, which is exactly the kind of commitment Caribbean governments have historically struggled to protect once the political cycle turns. Resilience and Sustainability is where I think the roadmap has the clearest, most immediate case: after Hurricane Melissa wiped out more than 40 percent of Jamaica's GDP in a single storm last October, AI-assisted early warning and recovery tools are not a future ambition. They are a budget decision waiting to be made.

The Number Nobody Said Out Loud in That Meeting

Here is the figure that should have been on every slide in Georgetown. Research compiled by the UN Economic Commission for Latin America and the Caribbean found that only 13 percent of Caribbean adults between 18 and 65 currently use generative AI tools, against roughly 55 percent of adults globally. That is not a small gap. That is the region using AI at roughly a quarter of the global rate, in 2026, three and a half years after ChatGPT put generative AI in front of the entire world at once.

The same research found that Caribbean AI companies made up less than 2 percent of all AI firms across Latin America and the Caribbean in 2023, with only about five startups in the entire CARICOM area whose core business is actually built on AI. Five. Across fifteen member states, a combined population of roughly 18 million people, and a diaspora many times that size. When StarApple AI opened its doors, it was not competing against a crowded field. It was one of the very few companies in the region doing the work at all, which is a fact I am proud of and also one I would much rather not still be citing this many years later.

These numbers matter more than the roadmap's four pillars, because they are the reason the roadmap exists in the first place, and they will still be true the day after the endorsement unless something specific changes. A policy roadmap does not move an adoption number. Funded training programs, affordable connectivity, local-language tools, and businesses that actually deploy AI into daily workflows move an adoption number. The roadmap can create the conditions for those things. It cannot substitute for them.

A Roadmap Is Not a Strategy, and Everyone in That Room Knows It

I want to be specific about the distinction, because it gets flattened in most coverage of this kind of announcement. A roadmap describes where you are trying to go and names the categories of work required to get there. A strategy assigns that work to specific ministries, with specific budgets, on specific timelines, measured against specific numbers. CARICOM now has the first. Almost none of its member states have the second. As of this month, Jamaica, Trinidad and Tobago, and Cuba are the CARICOM-adjacent territories actively developing national AI strategies. That leaves most of the region working from the regional roadmap alone, with no domesticated version of it and no assigned owner inside their own governments. For a country like Saint Lucia or Saint Vincent and the Grenadines, that is not necessarily a failure of will. Small island public services are stretched across everything from hurricane preparedness to debt servicing, and a fifteen-page regional framework does not staff itself into a national plan. It needs someone in each capital whose job that becomes.

This is where AI Guyana, based in the same city that just hosted COTED-ICT, has a genuine opportunity to press the advantage of proximity. The organization has been building local AI awareness and capacity in Guyana for years, and a government that just endorsed a regional roadmap in its own capital is about as receptive an audience as that kind of advocacy will ever find. The same argument applies to AI Jamaica, which has spent years pushing Jamaica's own AI conversation forward from the ground up, well before Jamaica's government began drafting a formal national strategy. Regional roadmaps get written in conference rooms. They get implemented by people who show up in ministries the following Tuesday and keep showing up. That second group is where the actual work happens, and it is smaller, in every country in this region, than it needs to be.

What Building the First Caribbean AI Company Taught Me About Waiting for Policy

I did not wait for a roadmap. Nobody in my position could have. When I started building AI capacity in Jamaica, there was no CARICOM framework, no COTED-ICT endorsement, no UNESCO consultation process to point to. There was a region with real problems, mostly no local AI talent pipeline to speak of, and a global technology moving fast enough that waiting for consensus meant falling further behind every month spent waiting. That experience is exactly why I do not dismiss this week's endorsement as theater, and exactly why I refuse to call it a finish line. Government frameworks matter enormously for the things individual founders and companies cannot do alone: setting data protection standards that businesses can build against with confidence, funding AI literacy in public schools at a scale no private company can match, and coordinating disaster-response AI systems across islands that share hurricanes but not governments. Those are genuinely government-shaped problems. But building the actual companies, training the actual workforce one cohort at a time, and shipping the actual products people use, that work does not wait for policy. It runs in parallel with it, or ahead of it, the way StarApple AI has run ahead of Caribbean AI policy for most of its existence.

What the Roadmap Gets Right, and It Is More Than Critics Will Admit

I have spent most of this article on what the roadmap does not do, so let me be fair about what it does. Four pillars that put AI literacy and resilience on the same level as governance is the right instinct; too many national AI strategies elsewhere treat education as an afterthought bolted onto a regulation document. Building the roadmap from over 1,000 consultations across two language communities means it was not drafted by a handful of officials in a single capital and then handed down. And endorsing it through COTED-ICT rather than leaving it as a UNESCO recommendation gives it a status inside CARICOM's own institutional machinery that a standalone UN document would never have had.

None of that is nothing. A region famous for producing well-written reports that quietly gather dust just produced a well-written report that fifteen governments formally agreed to own. That is a genuinely higher bar than the Caribbean has cleared on regional tech policy before.

The Ninety-Day Test

Here is what I will be watching between now and October. First, does any CARICOM member state publish a national implementation plan that names a lead ministry, a budget figure, and a deadline, rather than restating the roadmap's four pillars in different words. Second, does the Caribbean Development Bank or a comparable regional institution attach any funding mechanism to the roadmap, since smaller territories cannot implement AI governance and AI education programs out of general revenue they do not have. Third, does anyone in government put the 13 percent adoption figure on a dashboard anywhere and commit to reporting movement against it annually, because a number nobody tracks is a number nobody is accountable for.

If those three things happen, this week's endorsement will look, in hindsight, like the moment Caribbean AI policy stopped being fifteen separate improvisations and became one coordinated effort. If they do not happen by the time hurricane season peaks in September, the roadmap will join the long shelf of regional Caribbean agreements that were correct on paper and irrelevant in practice. I have built enough in this region to know which outcome the odds favor by default, and I have also spent fifteen years proving that default outcomes are not destiny. Grassroots education efforts are already doing exactly the kind of AI literacy work the Upskilling pillar describes, without waiting for a ministry to fund it first. The roadmap should be finding organizations like that and funding what they are already proving works, not commissioning a fresh pilot program to rediscover it.

CARICOM signed its name to the right document this week. Now every government that signed it gets to show, over the next ninety days, whether that signature was a commitment or a courtesy.

Frequently Asked Questions

What did CARICOM just endorse on Caribbean AI policy?

On July 7, 2026, the UNESCO Office for the Caribbean presented the Caribbean Artificial Intelligence Policy Roadmap to the 126th Special Meeting of CARICOM's Council for Trade and Economic Development, ICT configuration, known as COTED-ICT. Member states endorsed the roadmap, which was first launched in 2024 and was shaped by consultations with more than 1,000 people across the English and Dutch-speaking Caribbean. The roadmap is organized around four pillars: Culture and Creativity, Governance and Transformation, Upskilling and Education, and Resilience and Sustainability.

Does the CARICOM AI roadmap endorsement mean new laws or funding?

No. The COTED-ICT endorsement is a political signal that CARICOM member states agree on a shared framework for AI policy. It is not binding legislation, and it does not come with a regional budget attached. Turning the roadmap into national law, regulation, or funded programs is a separate step that individual governments have to take on their own timelines, which is exactly why the endorsement matters less than what governments do with it over the following months.

How many people in the Caribbean actually use AI?

According to research from the UN Economic Commission for Latin America and the Caribbean, only 13 percent of adults aged 18 to 65 in the Caribbean currently use generative AI tools, compared with roughly 55 percent of adults globally. The same research found that Caribbean AI companies made up less than 2 percent of all AI firms across Latin America and the Caribbean in 2023, with only about five startups in the CARICOM area whose core business is built on AI.

Which Caribbean countries have their own AI strategy?

As of mid-2026, Jamaica, Trinidad and Tobago, and Cuba are in the process of developing national AI strategies, working alongside the regional CARICOM framework rather than instead of it. Most other CARICOM member states do not yet have a published national AI strategy, which is part of why a shared regional roadmap, even a non-binding one, gives smaller territories a starting template they would otherwise have to build from nothing.

What are the four pillars of the CARICOM AI Policy Roadmap?

The roadmap is built on four pillars. Culture and Creativity addresses AI's effect on Caribbean cultural production, language, and creative industries. Governance and Transformation covers regulation, data protection, and public sector AI use. Upskilling and Education focuses on building AI literacy and technical skills across the region's workforce and education systems. Resilience and Sustainability addresses AI applications in disaster management, climate adaptation, and economic diversification, which matters directly to a hurricane-exposed region.

Why is Caribbean AI adoption so far behind the global average?

Research from ECLAC and the UN in the Caribbean points to gaps in investment, technical talent, and organizational readiness rather than a lack of interest or access. Most Caribbean businesses are still in early stages of general digital transformation, which makes AI adoption a second or third step rather than a first one. The region is not behind on access to AI tools, most of which are available for free or low cost globally. It is behind on the governance, training, and workflow redesign needed to turn access into actual use.

What should CARICOM governments do next after the roadmap endorsement?

The roadmap endorsement should be followed by three concrete steps within the next budget cycle: national implementation plans with named ministries and deadlines rather than general commitments, a regional funding mechanism through the Caribbean Development Bank or similar institutions to support smaller member states that cannot fund AI programs alone, and public reporting on the 13 percent adoption gap so progress against it can be measured year over year instead of left as an abstract policy goal.