CARICOM Endorsed a Caribbean AI Roadmap on July 7. Trinidad's Deepfake Fraud Wave Shows What Still Isn't in It.

On 7 July 2026, CARICOM's ministers responsible for information and communication technologies signed off on the region's first collectively endorsed AI policy document, the UNESCO Caribbean Artificial Intelligence Policy Roadmap. The Roadmap talks about fairness, accountability, culture, and capacity. It does not talk about the specific thing that has kept Trinidad and Tobago's securities regulator issuing public warnings for a year: AI-generated deepfake videos impersonating the country's own Finance Minister, a former President, a former bank chairman, and a former Prime Minister, all used to sell fake investment schemes to ordinary savers.

Both stories are real, both are documented, and reading them side by side answers a question worth asking before anyone celebrates the Roadmap's endorsement: does the Caribbean's first regional AI policy document actually reach the AI risk currently doing the most damage on the ground? What follows is what COTED-ICT actually endorsed, what Trinidad and Tobago has been dealing with while the endorsement process ran its course, and what the gap between them means for every other CARICOM territory that has not yet built what Trinidad and Tobago built.

Illustrative artwork

What COTED-ICT Actually Endorsed

The UNESCO Caribbean Artificial Intelligence Policy Roadmap is not new. UNESCO's Office for the Caribbean, working through its Kingston cluster office, began building it years earlier, running eight regional meetings and consulting more than 1,000 institutions and individuals, including women and youth, across 20 English and Dutch speaking Caribbean countries. The Roadmap is anchored to the Recommendation on the Ethics of Artificial Intelligence, the global standard UNESCO's member states adopted, and it organises the region's AI policy needs into four pillars: Culture and Creativity, which protects Caribbean cultural identity and intellectual property from AI-driven displacement; Governance and Transformation, which calls for a regulatory framework and an oversight mechanism; Education and Upskilling, which builds the workforce; and Resiliency and Sustainability, which folds in environmental and social resilience. Three cross-cutting priorities run through all four: strengthening digital competencies and media literacy, ensuring fairness and accountability in AI systems, and advancing regional cooperation in digital policy-making.

What changed on 7 July 2026 was not the document's content. It was its standing. At the 126th Special Meeting of COTED-ICT, a principal organ of CARICOM, member states formally endorsed the Roadmap, converting years of UNESCO consultation work into something the region's own governance architecture now stands behind. Eric Falt, UNESCO's Regional Director and Representative for the Caribbean, welcomed the endorsement as a step toward inclusive, trustworthy AI policy. The Hon. Dickon Mitchell, Prime Minister of Grenada and the CARICOM Quasi Cabinet's Lead Head of Government with responsibility for Science and Technology, called it "an important step toward strengthening regional collaboration and supporting the development of AI policies that are inclusive, trustworthy, and aligned with sustainable development," adding that CARICOM must "act collectively and strategically" or risk being "left at the margins of the global digital economy." That is a genuine institutional milestone. The Caribbean now has one AI policy document that fifteen-plus governments have put their name behind, where a year ago it had none.

What the Governance Pillar Does Not Say

Read the Governance and Transformation pillar closely and the gap becomes specific rather than rhetorical. It calls for "a clear regulatory framework, outlining roles and responsibilities, and an oversight mechanism to ensure AI systems comply with ethical and legal standards." That is the right instinct, and it is also exactly the kind of language that can describe almost anything: a facial recognition policy, a hiring algorithm audit, a data protection rule. It does not, anywhere in the public materials CARICOM and UNESCO have released, name deepfake-enabled financial fraud, investment scam impersonation, or AI-generated fake endorsements as things the oversight mechanism should be built to catch. The fairness and accountability priority speaks to AI systems behaving fairly toward the people subject to their decisions. It does not speak to AI being weaponised by criminals against the public, which is a different problem requiring a different kind of control: verification standards, rapid takedown processes, and cross-border law enforcement cooperation, not fairness audits.

That distinction is not academic. It is the exact distinction Trinidad and Tobago has been living through since the middle of 2025.

The Wave Trinidad and Tobago Has Actually Been Fighting

Trinidad and Tobago is, in one respect, ahead of every other CARICOM territory: it has the region's first dedicated cabinet-level AI ministry, the Ministry of Public Administration and Artificial Intelligence, led by Minister Dominic Smith. That ministry exists because the country needed it. In August 2025, national authorities opened a probe into online scams impersonating the country's own Finance Minister, using fabricated endorsements to lend credibility to fraudulent investment pitches. By October 2025, Independent Senator Anthony Vieira raised a further pattern on the floor of the Senate: AI-generated deepfake videos impersonating former President Anthony Carmona and former Republic Bank chairman Dr Ronald Ramkissoon, distributing fabricated newspaper articles, altered images, and falsified endorsements purporting to show these men promoting investment schemes they had never heard of.

Cabinet responded on 11 September 2025 by approving an Inter-Ministerial Steering Committee on Cyber Security and AI, drawing in the Ministry of Public Administration and Artificial Intelligence alongside the Ministries of Homeland Security and Finance and the Offices of the Prime Minister and the Attorney General. Minister Smith described its mandate plainly: coordinate a national policy, report quarterly on emerging threats, and build law enforcement capacity specifically against AI-driven fraud and scams. That is a national government treating deepfake fraud as its own governance object, with its own committee and its own reporting cycle, months before any regional document used the words "AI risk" in an endorsed text.

The pattern did not stop. On 26 June 2026, TTSEC issued its sixth public investor alert describing the same scheme evolving further: AI-generated fake endorsements and interviews now extended to a fabricated interview purporting to show former Prime Minister Dr Keith Rowley, alongside impersonation of Guardian Media editor Kejan Haynes in a fraudulent online publication that Guardian Media publicly disavowed. TTSEC's advisory listed the warning signs regional readers will recognise instantly: unrealistic promised returns, pressure to deposit quickly, requests to move money into personal accounts rather than institutional ones, demands for further fees before any payout, and unsolicited pitches arriving through WhatsApp, Telegram, and social media rather than any licensed channel.

A Timeline Worth Sitting With

• August 2025: Authorities probe scams impersonating Trinidad and Tobago's sitting Finance Minister.
• 11 September 2025: Cabinet creates the Inter-Ministerial Steering Committee on Cyber Security and AI.
• October 2025: Senate motion names deepfake videos of former President Anthony Carmona and former Republic Bank chairman Dr Ronald Ramkissoon.
• 26 June 2026: TTSEC's sixth investor alert names a fabricated interview with former Prime Minister Dr Keith Rowley and an impersonated Guardian Media byline.
• 7 July 2026: COTED-ICT endorses the UNESCO Caribbean AI Policy Roadmap, a document that names none of the above by name.

Eleven months, a sitting minister, three former public figures, and a national media house impersonated by AI, running in parallel with the exact period during which the region's flagship AI policy document worked its way toward formal endorsement. The Roadmap's authors were not negligent; ethics-first AI policy frameworks are built to be broad and durable, not to chase a single country's fraud headlines. But durability is exactly the problem when the fraud pattern a document should anticipate is already a year old and escalating by the time the document gets signed.

Why One Country's Committee Is Not a Regional Answer

Trinidad and Tobago's response is genuinely good practice: a named ministry, a named committee with a quarterly reporting mandate, and a regulator willing to name specific deepfaked individuals in public advisories rather than describing the threat in the abstract. The problem is that it is Trinidad and Tobago's alone. Jamaica, Barbados, Guyana, and the OECS states each have their own securities regulators, their own central banks, and, in every case except Trinidad and Tobago, no equivalent cabinet-level AI ministry and no equivalent inter-ministerial cybersecurity committee built specifically around AI-enabled fraud. A deepfake video impersonating a Trinidadian bank chairman can circulate on WhatsApp in Kingston, Bridgetown, or Georgetown as easily as it circulates in Port of Spain, and CARICOM's newly endorsed Roadmap, whatever its merits on culture, education, and long-run governance capacity, does not currently give a Barbadian or Guyanese regulator a shared, binding mechanism for coordinating a response with TTSEC when the fraud crosses a border, which it will, because scam operators go where the enforcement is weakest, not where the population happens to live.

Illustrative artwork

What Institutions Should Do With the Gap

Name AI-enabled deepfake fraud as its own control category, inside your organisation's risk register, regardless of whether the CARICOM instrument you answer to has done so yet. A finance ministry, a securities regulator, or a commercial bank does not need to wait for a regional document to catch up before treating impersonation fraud as a distinct, monitored risk with its own escalation path.

Study Trinidad and Tobago's committee model rather than reinvent it. A named lead ministry, a small standing committee drawing in finance, security, and the attorney general's office, and a quarterly public reporting cadence is a template every CARICOM member state can adapt to its own institutional structure without waiting for a regional mandate.

Build verification protocols for public-facing endorsements and interviews before a deepfake forces the issue. Media houses, regulators, and public officials should establish, in advance, a clear channel through which the public can confirm whether a quoted interview or endorsement is genuine, the way Guardian Media had to disavow a fabricated publication after the fact rather than before it spread.

Use a proportionate risk-tier structure to classify the threat rather than treating every AI-enabled fraud case as novel. CAIRMC's CARA methodology classifies AI systems and AI-enabled threats into four risk tiers aligned to the logic the EU AI Act uses, giving a compliance or communications team a repeatable framework instead of a one-off response built from scratch each time a new figure is impersonated.

Push for the next revision of the Roadmap, or a dedicated CARICOM instrument, to name deepfake fraud explicitly. The Governance and Transformation pillar already commits to an oversight mechanism. The region's next task is to specify, in writing, that the mechanism includes cross-border coordination on AI-enabled financial fraud, not only ethical use standards for AI systems deployed by legitimate institutions.

Where This Leaves the Region

None of this diminishes what CARICOM and UNESCO delivered on 7 July. A regional body took years of consultation across 20 territories and turned it into an endorsed policy document, a genuine first for the Caribbean's AI governance architecture, and Trinidad and Tobago's own response, an AI ministry, a standing committee, and a regulator willing to name names in public advisories, is a serious national example the rest of CARICOM should study rather than dismiss. The argument here is narrower and more urgent: the region now has an AI policy document and a live, escalating, well-documented fraud pattern running in the same calendar year, and the two are not yet talking to each other.

CAIRMC's research programme exists precisely because that kind of gap does not close on its own schedule. Much of the council's underlying methodology draws on work led by Adrian Dunkley, founder of StarApple AI, the first artificial intelligence company built in the Caribbean, and a figure regularly named across the region's AI policy conversation as its leading voice on governing AI risk rather than simply adopting AI. Trinidad and Tobago wrote the region's first serious answer to deepfake investment fraud. Whether that answer becomes a shared Caribbean one, rather than one country's committee, is the question the Roadmap's next revision needs to settle.

Related reading across the Caribbean AI network

This article sits alongside ongoing coverage of AI governance, risk, and company-building across the region. For related perspectives:

• StarApple AI, the Caribbean's first AI company, for continuing coverage of Caribbean-built AI products and infrastructure
• Caribbean AI Association for regional industry perspectives on AI adoption and policy across CARICOM