Only 6% of Caribbean Boards Think They Watch AI Closely Enough. A Regional Framework Already Exists to Close That Gap.

Kingston, Jamaica. Photo by Russian Supreme on Unsplash.

A board that does not know which of its own systems make automated decisions cannot be said to govern AI, whatever line item sits in its risk register. PwC's 2026 Caribbean Corporate Governance Survey put a number on that gap this year: of 154 directors across six territories, only 6 percent believe their board spends enough time understanding AI's impact on the business, and only 9 percent believe they get enough information to manage the risk it creates. That is not a technology problem. It is a board agenda problem, and it has a fix that does not require CARICOM to invent anything new.

What just landed on the table

Three things arrived within nine weeks of each other. On 7 July 2026, CARICOM's Council for Trade and Economic Development formally endorsed the UNESCO Caribbean AI Policy Roadmap, built around four pillars: culture and creativity, governance and transformation, education, and resilience and sustainability. On 23 July, the Caribbean Telecommunications Union's Caribbean AI Task Force, more than 35 members drawn from governments, regulators, academia and industry over a one-year mandate, launched its final report, Toward Harmonized AI Policies and Recommendations for the Caribbean, at the first Caribbean Artificial Intelligence Forum. And on 22 September, attorney Maralyn Ballantyne wrote in Searchlight that AI is "already spiralling out of control" in the region and that CARICOM needs one coordinated response rather than fifteen separate ones.

Read in sequence, those three events describe a region that has done the policy work and is now waiting for someone to act on it. The roadmap and the task force report both talk about national and regional coordination. Neither one tells a bank, an insurer or a ministry what its own board should be asking on Tuesday. That is where the PwC numbers matter: they measure exactly the layer the policy documents leave open.

What the numbers actually say

PwC surveyed directors between November 2025 and January 2026 across The Bahamas, Barbados, Grenada, Jamaica, Saint Lucia and Trinidad and Tobago, with Bermuda included for the first time. Alongside the 6 percent and 9 percent figures, the survey found that most boards say they lack the time, the skills and the training to oversee AI properly, not the willingness. PwC's own framing of the finding was blunt: recognition alone is not enough, and Caribbean boards now have to convert awareness into action.

That gap sits next to a separate, widely cited figure on the demand side: generative AI use among Caribbean adults runs around 13 percent against a roughly 55 percent global average, a spread I have written about before. Put the two numbers together and the shape of the problem changes. This is not primarily a region moving too slowly to adopt AI. It is a region where staff and business units are already using AI tools, and the governance layer meant to know what those tools are doing has not caught up. Adoption outrunning oversight is the more dangerous order of the two.

Why I have been arguing this before the survey existed

I founded StarApple AI in Kingston in 2016 because the Caribbean could not afford to be a passive buyer of AI systems built for other markets. Governance was the same argument from a different angle, which is why I chair the Caribbean AI Risk Management Council and serve as President of the Caribbean AI Association. A risk framework written for a US bank assumes a compliance department, a chief risk officer with an AI mandate, and vendors who will answer a detailed audit questionnaire. Most Caribbean boardrooms have none of those three, and an imported framework fails here for that reason even when it is well written.

CAIRMC's published guidance covers model risk for Caribbean banks, third-party and vendor AI risk, and the board governance gap directly, mapped against the EU AI Act and NIST's AI Risk Management Framework, not translated line for line from either. That mapping work exists because a director in Bridgetown or Port of Spain asking "does this apply to us" deserves an answer specific to a Caribbean regulatory environment, not a US or European one with the names changed.

A concrete example of what happens without it

In 2026, Trinidad and Tobago's finance minister was targeted twice by fabricated video and audio, part of a broader wave of AI-generated fraud and impersonation aimed at the country's political and financial figures. Trinidad's securities regulator issued public alerts warning investors and the public not to trust unverified video of officials making financial claims. I wrote about the wave in detail elsewhere on this site, and the detection problem behind it is real: models trained mostly on North American and European faces and accents perform worse on Caribbean material, the kind of clip circulating on WhatsApp in Port of Spain.

The lesson for boards is not about deepfakes specifically. It is that the institutions most exposed, banks, insurers, brokerages, are the ones whose customers rely on trusting a video, a voice note or a call from someone claiming to represent them. A board that has never asked what its own fraud detection and customer verification systems can and cannot catch has no basis for reassuring anyone when the next fake message names its company.

What this means for the region

The policy layer is in reasonable shape. CARICOM has an endorsed roadmap, the CTU has a detailed task force report, and UNESCO has convened ministers and regulators on the ethics side for three years running. What is missing sits below all of that, at the level of a specific board agenda in a specific quarter.

Five things close the gap, and none of them require CARICOM to pass a new law first.

1. Boards: put AI on the agenda as a standing item, not a briefing. Ask management for an inventory of every system that makes or materially informs a decision about a customer, an employee or a transaction, with a named owner for each one.
2. Risk committees: ask for AI reporting on the same cadence as credit or cyber risk. An annual AI briefing is not oversight of a technology that changes every few months.
3. Regulators: publish a short, sector-specific checklist before writing a full AI law. A one-page expectation for banks and insurers this quarter does more than a comprehensive statute two years from now.
4. CARICOM: assign the endorsed roadmap to a named body with a budget line. Model legislation that member states can adapt moves faster than fifteen governments each starting from the same set of principles and drafting separately.
5. Smaller businesses without a formal board: do the one-page version. List every AI tool in use, who is responsible for it, and what happens if it is wrong. The accountability question does not go away because there is no boardroom to ask it in.

Where the argument is weakest

A framework, however well mapped to Caribbean regulation, only works if a board actually uses it, and PwC's finding is precisely that most do not yet have the skills in the room to use one well. Training closes that gap slower than any of us would like, and I do not have a shortcut for it. The realistic next milestone is not full compliance by year end; it is a first AI inventory on a board agenda, which is a smaller thing than it sounds and still further than most Caribbean boards have gone.

Frequently Asked Questions

What did PwC's 2026 Caribbean Corporate Governance Survey actually find about AI?

PwC surveyed 154 board directors between November 2025 and January 2026 across The Bahamas, Barbados, Grenada, Jamaica, Saint Lucia and Trinidad and Tobago. Only 6 percent believed their board spends enough time understanding the impact of AI on the business, and only 9 percent believed they receive enough information to address the risks it creates.

Is there already a regional AI governance framework Caribbean boards can use?

Two pieces exist. CARICOM's Council for Trade and Economic Development endorsed the UNESCO Caribbean AI Policy Roadmap on 7 July 2026, and the Caribbean Telecommunications Union's Caribbean AI Task Force published its final report, Toward Harmonized AI Policies and Recommendations for the Caribbean, at the first Caribbean Artificial Intelligence Forum on 23 July 2026. Neither is a board-level control framework on its own, which is the gap a body such as the Caribbean AI Risk Management Council exists to close.

What does a board actually need to ask about AI risk?

Start with an inventory: which systems in the business already use AI, who approved them, and who is accountable if one fails. Then ask for the same reporting cadence given to credit risk or cyber risk, not a once-a-year briefing. A board that cannot name its five highest-risk AI use cases is not overseeing AI, whatever the minutes say.

What happened with the deepfakes of Trinidad and Tobago's leaders?

In 2026, Trinidad and Tobago's finance minister was targeted by fabricated video and audio twice, part of a wider wave of AI-generated fraud and impersonation content aimed at Caribbean political and financial figures. The country's securities regulator issued public alerts, and it became a central example in the region's push for a shared verification framework and CARICOM's own AI roadmap discussions.

Why do Caribbean boards lag on AI oversight when adoption is already happening inside their companies?

Staff and business units adopt AI tools faster than governance catches up, which is normal everywhere, but it is sharper in the Caribbean because most boards here are smaller, meet less often, and rarely have a director with direct AI experience. The result is oversight built for a technology that changes every few months being handled at a pace built for technology that changes every few years.

Does this apply to small and medium-sized Caribbean businesses, or only large listed companies?

PwC's sample was board directors, so it skews toward listed and larger regulated companies. Smaller businesses face the same exposure without a board at all: an owner using an AI tool for hiring, pricing or customer service carries the same accountability question with nobody else in the room to ask it. The fix scales down to a simple written review of what AI tools the business uses and who is responsible for each one.

Where can a Caribbean company get an AI governance framework built for the region instead of adapting a US or EU one?

The Caribbean AI Risk Management Council publishes region-specific guidance covering model risk for Caribbean banks, vendor and third-party AI risk, and board governance gaps, mapped against international standards such as the EU AI Act and NIST's AI Risk Management Framework rather than a straight copy of either. The Caribbean AI Association is the broader forum for regional coordination and advocacy on the policy side.

What should CARICOM do next on AI governance?

Assign a named regional body with a budget line to turn the endorsed roadmap into model legislation that national governments can adapt, rather than leaving each of fifteen member states to draft its own AI law from a shared set of principles. Regional coordination on financial services and telecoms shows the model works; AI governance needs the same treatment.