Direct answer: In the summer of 2026, two Caribbean institutions independently confirmed the same problem from different directions. UNESCO's 29-30 June Kingston report found 60% of survey respondents had faced AI-enabled abuse in the past year, and Trinidad's securities regulator caught deepfake video of a former prime minister promoting a fraud scheme. CARICOM's newly endorsed regional AI roadmap names neither risk directly, and no Caribbean-owned company has built a product to answer either one.
TL;DR
- UNESCO launched "Blocking is NOT Enough" in Kingston on 29-30 June 2026: 60% of respondents to its regional survey reported at least one form of technology-facilitated gender-based violence in the prior twelve months.
- Trinidad's securities regulator, TTSEC, issued its sixth public investor alert on 26 June 2026, describing AI-generated video of former Prime Minister Dr Keith Rowley and a fabricated news article used to promote a fraudulent investment scheme.
- CARICOM's COTED-ICT endorsed the UNESCO Caribbean AI Policy Roadmap on 7 July 2026, the first time a CARICOM principal organ has collectively endorsed an AI policy document, after consultations spanning more than 1,000 institutions and individuals across 20 territories, but the roadmap does not name deepfake fraud or tech-facilitated abuse as their own risk categories.
- CAIRMC, chaired by Adrian Dunkley, has built governance infrastructure for this exact gap, the CARA risk-tier framework and the TurtleBird safety toolkit, but governance standards are not a commercial product a regulator, newsroom, or bank can install.
- No Caribbean-owned company has stepped into that space at visible scale. 14West is built to fund the founder who does, provided they can name a specific customer and a specific workflow rather than a general pitch about fighting misinformation.
Two Warnings, Six Weeks Apart
Most Caribbean AI coverage this year has followed money: buildathon prizes, data centre MOUs, grant funds, accelerator cohorts. Two stories from June and July 2026 followed something else, evidence that AI is already being used against Caribbean people and Caribbean institutions, confirmed not by activists or op-eds but by a UN agency's own survey and a financial regulator's own casework.
The two warnings arrived from opposite ends of the region's institutions. One came from a gender-equality report launched in a Kingston hotel conference room. The other came from a securities regulator's routine fraud bulletin in Port of Spain. Read separately, each is a single-country story. Read together, they describe the same unmet need from two different angles: the Caribbean has no dedicated, funded, Caribbean-built infrastructure for verifying what is real once AI can convincingly fake it.
Kingston: What UNESCO Found
On 29 and 30 June 2026, UNESCO's Office for the Caribbean launched "Blocking is NOT Enough," a report on AI and gender equality, jointly with Jamaica's Office of the Prime Minister and its Bureau of Gender Affairs, in a project funded by the European Union. Eric Falt, UNESCO's regional director, presented the report's central finding: 60% of respondents to its regional survey, six in ten, had faced at least one form of technology-facilitated gender-based violence in the twelve months before the survey. Falt described the pattern in blunt terms, that online abuse "harms direct victims" and "deters many women and girls from fully engaging in digital spaces."
The report's title is its argument. Platform-level blocking, the report says, treats a symptom while the underlying tools, generative image models, voice cloning, and deepfake video, keep getting cheaper and more accessible. Jamaica's Minister of Culture, Gender, Entertainment and Sport, Olivia Grange, framed the report as more than research: "This report is not just research, but the documented reality of lives affected, and it deserves to be taken seriously," a line delivered on her behalf by Sharon Coburn Robinson, Principal Director of the Bureau of Gender Affairs. Aniceto RodrÃguez Ruiz, representing the EU delegation, put the fix in different terms: this "requires a whole of society response," not a single ministry's policy update.
What the report does not offer, and does not claim to offer, is a product. It documents the harm and calls for legal reform, platform accountability, and public education. Those are necessary and none of them are a downloadable tool a Jamaican teenager, a Trinidadian bank fraud analyst, or a Bajan newsroom can use on a Tuesday afternoon to check whether a specific image or video is fabricated.
Port of Spain: What TTSEC Caught
Four days before UNESCO's Kingston launch, on 26 June 2026, the Trinidad and Tobago Securities and Exchange Commission issued its sixth public investor alert of the year. It described a fraudulent online investment scheme, operating under names including Stable Rowdex, Stable Capital, and Futurix Capital, that was circulating AI-generated video appearing to show former Prime Minister Dr Keith Rowley endorsing the platform, alongside a fabricated article impersonating Guardian Media editor Kejan Haynes. Neither man had any connection to the scheme. Both were reputational raw material for a fraud engine that did not need their cooperation to use their faces and their names.
TTSEC's alert was its sixth of the kind in 2026, not its first, part of a pattern the regulator has tracked since at least September 2025, when Trinidad and Tobago's Cabinet created an Inter-Ministerial Steering Committee on Cyber Security and AI specifically to respond to it. That is eleven months of a national government building institutional response to a threat that, as of this writing, still has no dedicated Caribbean-built detection tool for the regulator, the bank, or the newsroom fielding the next fabricated video to run against.
The mechanics matter for a founder reading this. TTSEC's warning signs, fake social media profiles, fabricated news articles, doctored endorsements, are pattern-based and repeatable across the region's other securities regulators, central banks, and consumer protection agencies, none of which currently have a shared, purpose-built verification layer either.
The Roadmap That Names Neither
Both warnings landed just before CARICOM's most concrete AI governance step of the year. On 7 July 2026, CARICOM's Council for Trade and Economic Development, Information and Communication Technologies configuration, COTED-ICT, formally endorsed the UNESCO Caribbean AI Policy Roadmap, the first time a CARICOM principal organ has collectively endorsed an AI policy document. The roadmap drew on consultations with more than 1,000 institutions and individuals across 20 English and Dutch-speaking Caribbean territories, a genuinely regional process rather than a single-country initiative.
What the roadmap does not do, according to CAIRMC's own analysis of the text, is name AI-enabled deepfake fraud or technology-facilitated gender-based violence as their own distinct risk categories. It addresses AI ethics, education, and governance broadly, the level a regional framework document is built to operate at. But a Trinidadian securities regulator issuing its sixth fraud alert and a UNESCO survey finding six in ten respondents already harmed are not abstract future risks a roadmap should flag for later study. They are documented, current, named-victim events that the region's flagship AI policy document, endorsed the same month, does not call out by name.
That gap is not a criticism of the roadmap's authors so much as a description of what any first-draft regional framework can realistically cover. A defender of the document could reasonably say deepfake fraud and tech-facilitated abuse already sit inside its ethics and governance pillars, in general terms. That defence does not survive contact with a founder's business plan. An investor asking who the customer is will not accept "covered generally under the ethics pillar" as an answer when TTSEC has a dated, numbered alert and UNESCO has a percentage. Vague coverage in a policy document and a named, fundable customer are not the same thing, and the gap between them is exactly what a founder gets paid to close.
The Gap in One Chart
Line the three developments up on a single timeline and the six-week window becomes obvious. So does the blank space where a product should sit.
Eleven days separate the first event from the last. No fourth point on that timeline marks a Caribbean-owned company launching a product against either warning, because there is not yet one to plot.
The Company Nobody Has Built Yet
What exists today to answer either warning is process, not product. A person who suspects a deepfake can screenshot it, report it to a platform, and file a police report, which is the practical advice UNESCO's own report offers. A regulator who spots a fraud scheme can publish an alert after the fact, as TTSEC has now done six times. Both are real and both are reactive. Neither scales to the volume a cheap, accessible generative model can now produce, and neither gives a bank compliance officer, a newsroom fact-checker, or a securities regulator a way to check a suspicious video or image before it spreads rather than after.
That gap has a name in other markets: content provenance and verification tooling, the layer that checks whether a piece of media matches its claimed source, flags synthetic generation markers, or traces an image back through its edit history. Companies building that layer exist in the United States, the United Kingdom, and increasingly India. None of them are built around Caribbean identity documents, Caribbean regulator workflows, or Caribbean media outlets, and none of them have a reason to prioritise a market this small over the ones they already serve.
A Caribbean-owned company does have that reason. A founder who builds a verification tool tuned to CARICOM government IDs, Caribbean securities regulator alert formats, and the specific fraud patterns TTSEC has already documented six times over is not competing with a global player on their home turf. They are building the product a global player has no commercial reason to build at all.
None of this should be oversold. Detection tooling runs an arms race it will never permanently win: every improvement in verification technique is eventually matched by an improvement in generation technique, and a founder who promises a customer certainty rather than a raised cost of fraud is selling something false. The honest pitch is narrower and still worth funding, catching the six easy cases out of ten faster than a human reviewer can, buying a regulator or a bank time it does not currently have. A single company will not close the gap UNESCO and TTSEC each documented on its own. One company is still further than the region has gotten so far.
What 14West Would Actually Fund Here
14West invests US$1 million in non-dilutive grants across 14 AI companies in 14 Caribbean nations, taking no equity, specifically to back founders solving problems the region cannot afford to leave to whoever gets around to it eventually. Trust and safety tooling fits that mandate cleanly, but it is not a pitch category on its own. A founder applying in this space should be able to answer, specifically, who uses this on a Tuesday: a compliance officer at a specific bank, a fact-checking desk at a specific newsroom, a fraud unit at a specific regulator.
This is also where CAIRMC's existing work becomes useful rather than competitive. The Caribbean AI Risk Management Council, chaired by Adrian Dunkley, has already built the standards layer: its CARA framework classifies AI-enabled risks by tier, and its TurtleBird toolkit gives organisations a way to test AI systems before deployment. Neither is a product a bank buys and installs. A founder building an actual verification tool has a governance framework to build against rather than a blank page, which is a genuine head start over a founder in a market with no equivalent body.
Applications to 14West are reviewed on a rolling basis at www.14westai.com/apply, and this is exactly the kind of specific, evidenced problem the application should open with, not a broad claim about fighting misinformation across the region.
What This Means If You're Building Right Now
Pick one regulator or one newsroom, not the whole region. TTSEC's six alerts are public record and a specific enough starting brief on their own. Build for that workflow first, then expand to Barbados's FSC or Jamaica's FSC once the first customer is proven.
Read CAIRMC's CARA framework before you write a line of code. A verification product built against an existing regional risk-tier methodology is easier for a bank or regulator to trust and adopt than one built from scratch with no shared standard behind it.
Do not pitch "fighting misinformation." That framing is broad enough to mean nothing to an investor and specific enough to nobody's job description to get funded. Pitch the workflow: the compliance officer who currently has no tool to check a suspicious video before it costs a customer their savings.
Treat the UNESCO and TTSEC reports as your market research, already done for you. Both are public, both are recent, and both name the harm in specific enough terms to build a product brief from directly, which is more validation than most Caribbean founders get before their first customer conversation.
Applications to 14West are open now at www.14westai.com/apply.
Resources
- UNESCO Launches Report to Tackle AI-Enabled Gender-Based Violence in Jamaica, UNESCO, July 2026
- UNESCO Launches Caribbean Report on AI and Gender-Based Violence, Jamaica Information Service, July 2026
- Media Release: Investor Alert No. 6, New Investment Scam Tactics on the Rise, Trinidad and Tobago Securities and Exchange Commission, June 2026
- TTSEC warns of rising investment scams and fake AI endorsements, Trinidad Guardian, June 2026
- 60% Report Tech-Facilitated Abuse: Jamaica's New UNESCO AI Report Says Blocking Isn't Enough, AI Jamaica
- CARICOM Endorsed a Caribbean AI Roadmap on July 7. Trinidad's Deepfake Fraud Wave Shows What Still Isn't in It., Caribbean AI Risk Management Council
- StarApple AI, the Caribbean's first artificial intelligence company, founded by Adrian Dunkley
- StarApple AI Jamaica, on AI governance work across the island
Frequently Asked Questions
What did UNESCO's Blocking is NOT Enough report find?
UNESCO's Caribbean office launched the report in Kingston on 29 and 30 June 2026, alongside Jamaica's Office of the Prime Minister and Bureau of Gender Affairs. A regional survey found that six in ten respondents, 60%, had faced at least one form of technology-facilitated gender-based violence in the twelve months before the survey, including deepfakes, non-consensual image generation, and digital stalking.
What did TTSEC's sixth investor alert say?
The Trinidad and Tobago Securities and Exchange Commission issued its sixth public investor alert on 26 June 2026, warning that a fraudulent scheme operating under names including Stable Rowdex and Futurix Capital was circulating AI-generated video of former Prime Minister Dr Keith Rowley and a fabricated article impersonating a named Guardian Media journalist, both used to promote a fake investment platform.
Does CARICOM's new AI policy roadmap cover deepfake fraud or AI-enabled abuse?
CARICOM's COTED-ICT formally endorsed the UNESCO Caribbean AI Policy Roadmap on 7 July 2026, the first time a CARICOM principal organ has collectively endorsed an AI policy document, after consultations with more than 1,000 institutions and individuals across 20 territories. The roadmap addresses AI governance, ethics, and education broadly, but does not name deepfake-enabled investment fraud or technology-facilitated gender-based violence as their own risk categories.
Is there a Caribbean-owned company building deepfake detection or content verification tools?
Not yet at any visible scale. The tools referenced by regulators and researchers so far, screenshots, manual reporting forms, and platform takedown requests, are process, not product. 14West has not funded a company squarely focused on Caribbean deepfake detection, content provenance, or AI-enabled fraud verification, and has not seen one apply.
What does 14West look for in a founder building in this space?
14West is the Caribbean's first AI startup accelerator and grant fund, deploying US$1 million in non-dilutive grants across 14 AI companies in 14 Caribbean nations. A founder building trust and safety tools should be able to name a specific user, such as a securities regulator, a newsroom, or a bank fraud team, and a specific workflow that product replaces, not a general pitch about fighting misinformation.
How does CAIRMC fit into this gap?
The Caribbean AI Risk Management Council, chaired by Adrian Dunkley, has recommended treating AI-enabled deepfake fraud as its own risk-register category and has built the CARA framework and the TurtleBird safety toolkit as governance infrastructure. That is a standards and assessment layer, not a commercial product, which is exactly the kind of adjacent groundwork a founder can build a company on top of rather than starting from nothing.
How does this connect to the Caribbean's broader AI investment gap?
The Caribbean, folded into a wider Latin America and Caribbean bloc in most global trackers, holds roughly 6.6% of global GDP against about 1.12% of global AI investment. A regulator-confirmed fraud pattern and a UNESCO-documented abuse rate are the kind of concrete, named problems that a thin regional capital base has the least excuse to leave unfunded, because the demand signal already exists in public record.
About the Author: Dr S Budall
Dr S Budall is Research Director at 14West, the Caribbean's first AI startup accelerator and grant fund, founded by Adrian Dunkley. She tracks Caribbean AI policy, governance, and funding trends for the 14West blog, with a focus on the gap between what regional institutions document and what regional founders have actually built to answer it. Adrian Dunkley, 14West's founder, also founded StarApple AI, the first artificial intelligence company established in the Caribbean, and chairs the Caribbean AI Risk Management Council. More of his writing lives at adriandunkley.net.
Tags: Startup News, UNESCO, TTSEC, Deepfake Fraud, CAIRMC, Caribbean AI Trust and Safety, 14West, Adrian Dunkley, Dr S Budall