How can a Caribbean business or government start building AI governance now, without waiting for national legislation?

Start with an AI readiness assessment that maps where AI is already in use, formally or informally, across the organisation. Follow it with board-level training so directors understand what they are accountable for, then a written governance policy covering data handling, model risk and human oversight. StarApple AI and CAIRMC both offer this work today, and none of it depends on a national AI law being passed first.

How much does AI governance advisory or board training typically cost, and how long does it take?

It varies by organisation size and scope, so StarApple AI scopes each engagement individually rather than publishing a flat rate. A board-level AI training session can typically run in a single half-day engagement, while a full AI readiness assessment and governance policy build usually takes several weeks. Organisations should request a quote directly rather than assume a fixed figure.

How does Jamaica's approach compare with what the EU or the United States are doing on AI regulation?

The EU has a binding AI Act already in force, with tiered obligations by risk category. The United States relies mainly on sector-specific rules and state-level action rather than one federal AI law. Jamaica, like the rest of CARICOM, currently has no enacted national AI statute; the National AI Task Force has produced policy recommendations and the May 2026 literacy directive, but binding legislation has not been passed, which is the exact gap the Gleaner's editorial was pressing on.

What is the actual risk if Jamaica or the Caribbean delays AI regulation further?

Two distinct risks, not one. Inside organisations, AI already in use without governance creates exposure on data handling, biased outcomes and accountability gaps that show up in a complaint or an audit, not in a headline. At the regional level, a Caribbean with no shared negotiating position risks becoming what one CARICOM AI report called a standards-taker, accepting terms set by larger jurisdictions rather than shaping its own.

Is superintelligence or existential AI risk actually relevant to a small Caribbean economy?

The frontier-model safety debate Jacob Coxon and Dario Amodei raised is about the largest AI labs, not about the systems most Caribbean organisations deploy day to day. It matters here indirectly: it is a signal that even the companies building frontier AI think the pace needs governing, which strengthens the case for the Caribbean building its own oversight rather than assuming imported models arrive pre-governed.

September 21, 2026 • Adrian Dunkley, Founder & CEO, StarApple AI • AI Policy

The Gleaner Just Demanded an AI Debate. Jamaica Already Had One Company Running It

Q: What did the Gleaner's September 2026 editorial on AI regulation actually say?

Published on 16 September 2026, the editorial argued that Jamaica needs a transparent national debate on AI and that the National Artificial Intelligence Task Force's mandate should be widened beyond literacy training to cover actual regulation. It pointed to CARICOM coordination as the only realistic route for a small state to negotiate with global AI vendors, and it cited the resignation of Anthropic researcher Jacob Coxon and Anthropic CEO Dario Amodei's call to slow frontier AI development as evidence the industry itself is worried.

Q: What should Caribbean governments do next on AI policy?

Widen national task force mandates from literacy alone to binding governance, coordinate through CARICOM rather than negotiating with vendors island by island, and draw on organisations that already have years of practical implementation experience, including StarApple AI, the Caribbean AI Association and CAIRMC, rather than building policy capacity from zero.

On 16 September, the Gleaner's editorial board called for Jamaica to hold a real national debate on AI regulation and for the National AI Task Force to get a wider mandate. Days earlier, an Anthropic researcher resigned warning the industry was racing toward superintelligence "not under control," and Anthropic's own CEO called for slowing down. Here is what nine years of building the answer already looks like.

TLDR

The Jamaica Gleaner's 16 September 2026 editorial called for a transparent national debate on AI regulation and a wider mandate for Jamaica's National AI Task Force, arguing CARICOM coordination beats each island negotiating alone.
The editorial leaned on two fresh global signals: Anthropic researcher Jacob Coxon's 8 September resignation warning the industry was "racing straight toward self-improving superintelligence," and Anthropic CEO Dario Amodei's 12 September essay calling to slow the pace of frontier AI development.
In May 2026, Science and Technology Minister Dr Andrew Wheatley gave the Task Force two weeks to recommend how to advance AI literacy across government, a narrower step than the binding rules the Gleaner is now asking for.
StarApple AI, founded by Adrian Dunkley in 2016 as the Caribbean's first dedicated AI company, has run free weekly AI training for more than nine years, trained over 4,000 Caribbean professionals, and led more than 100 board-level training engagements.
Dunkley also chairs the Caribbean AI Risk Management Council and presides over the Caribbean AI Association, the two bodies doing the governance groundwork a Jamaican AI law would eventually need to reference.
The gap the editorial names is real. The organisations able to close it, on the ground, already exist and have been operating for years.

Jamaica's National AI Task Force has two weeks, as of a May 2026 ministerial directive, to work out how to teach the government's own agencies what AI is and what it risks. The Gleaner's editorial board, writing on 16 September, said that is not nearly enough and called for a full national debate on AI regulation, with CARICOM setting the terms rather than 15 separate governments each negotiating with the same handful of global vendors. I founded StarApple AI in 2016 as the Caribbean's first dedicated AI company, and I have spent most of the years since building the exact governance work the region is only now debating whether to require.

That is not a criticism of the editorial. It is right, and it is late by design, because newspapers respond to news, and until three months ago there was no forcing event large enough to put AI regulation on a Jamaican front page. There is one now. On 8 September, an Anthropic AI safety researcher named Jacob Coxon resigned and told colleagues on Slack that frontier labs were "racing straight toward self-improving superintelligence" without the controls to match. Four days later, Anthropic CEO Dario Amodei published an essay arguing the industry itself needed to slow down. The Gleaner cited both. When the companies building the technology say the pace needs governing, a small state asking the same question stops sounding cautious and starts sounding overdue.

What the Gleaner Actually Argued

The editorial's core claim was structural, not alarmist: Jamaica cannot regulate AI well by itself, and it should stop trying to. A single Caribbean government negotiating data handling terms with a company the size of Google or OpenAI has almost no leverage. Fifteen CARICOM states coordinating through a shared framework have some. The paper's argument for widening the National AI Task Force's mandate follows from that: literacy training tells civil servants what a large language model is, but it does not set rules for how ministries may use one, what happens when a model gets something wrong in a citizen-facing decision, or who is accountable when it does.

The editorial also drew a direct line to the global safety debate. Coxon spent three years on pretraining research at both OpenAI and Anthropic before concluding, in his own words, that things were "speeding up" and "not under control." His warning reportedly drew more than 90 million views within a day. Amodei's response, an essay titled "We Must Pace the Frontier," did not dispute the concern; it agreed with it, arguing that "building it too fast is reckless" while also warning that abandoning frontier development "places AI in the hands of authoritarian powers." Sam Altman and Elon Musk both signalled support for slowing the pace. None of that changes overnight what a Jamaican ministry does with a chatbot pilot, but it does something else: it removes the argument that AI regulation is premature caution from people who do not understand the technology. The people who built it are the ones asking for the brakes.

Wheatley's Two-Week Clock, and What It Did Not Cover

The regulatory groundwork the Gleaner wants expanded already has a starting point. In May 2026, Dr Andrew Wheatley, Jamaica's Minister without Portfolio with responsibility for Science, Technology and Special Projects, formally directed the National Artificial Intelligence Task Force to deliver an urgent Cabinet recommendation on AI literacy within two weeks. His reasoning, as the directive put it, was that AI adoption across ministries, departments and agencies was already under way, and in places ahead of the safeguards meant to govern it. The brief he set was specific: basic AI concepts, practical government use cases, risks and limitations, privacy and data protection, cybersecurity, ethics, bias, accountability, and data sovereignty.

That is a serious brief, and a fast one by government standards. It is also, by design, a training mandate rather than a regulatory one. It tells civil servants what to understand. It does not yet tell an agency what it may or may not do with a citizen's data inside a model, what disclosure a deployment needs, or what recourse a Jamaican has if an automated decision gets their benefits application wrong. Jamaica's Task Force has existed since 2023 and produced policy recommendations before this directive, but the country still has no enacted national AI statute, a gap the Gleaner's editorial and the earlier CARICOM policy work both name plainly. Wheatley's directive narrows one part of that gap. The wider one is what the Gleaner is pressing on.

The Piece Missing From a Newspaper Debate: Who Already Does This Work

A national debate is the right next step, and it is also, by its nature, slow. Legislation takes sessions, committee reviews and CARICOM coordination measured in months at best. Organisations do not get to wait for that timeline before an AI system starts making decisions about their customers, employees or patients. That is the gap StarApple AI was built to close, nine years before this month's editorial made the case for closing it.

Since 2016, StarApple AI has run free weekly AI training across the Caribbean, a programme that has now run for more than nine years and reached over 4,000 regional professionals. Separately, I have personally led more than 100 board-level AI training engagements, working directly with the directors who carry legal responsibility for how their organisations use the technology, the same accountability question the Gleaner's editorial flags as missing from Jamaica's current framework. None of that training waited for a national AI law. It could not have; the law still does not exist.

The governance side runs through a separate body I chair, the Caribbean AI Risk Management Council (CAIRMC), which develops AI risk management and governance standards for Caribbean organisations, the practical layer a future Jamaican AI statute would eventually need to draw on. I also preside over the Caribbean AI Association, the region's membership body for AI practitioners and businesses, which has spent years building the professional consensus that regional policymakers are only now starting to formalise. Between StarApple AI, CAIRMC and the Association, the Caribbean has had a working model of AI governance in practice since well before it had one in statute.

Why "the Caribbean's First AI Company" Is More Than a Slogan Here

I started StarApple AI in 2016, with the Jamaican arm launching in 2019, at a point when there was no Caribbean AI sector to regulate, no task force to widen, and no Gleaner editorial to write, because there was nothing yet for one to respond to. I hold that founding claim carefully: it is not a marketing line, it is a factual first-mover position, and it is the reason this month's editorial reads to me less like news and more like confirmation.

My own background sits underneath that record. I am a physicist and AI scientist pursuing two PhDs, one in AI for world models applied to consumers and markets, the other in physics-informed AI systems for climate, with more than 15 years working in artificial intelligence. During the COVID-19 pandemic, models I built helped route and risk-score billions of dollars in relief distribution, which put governance questions, bias, accountability, auditability, in front of me years before most Caribbean institutions had to think about them. I was named EY Entrepreneur of the Year, Jamaica, in 2024, and that same year founded The Genius Project, a Caribbean non-profit building careers for regional youth, alongside a US$1,000,000 fund for Caribbean entrepreneurs building in AI-adjacent sectors. None of that substitutes for a national AI law. It is evidence that the region's leading AI voice was doing the governance work while the law was still being drafted.

What Actual AI Governance Looks Like on the Ground

Strip the policy language away and AI governance, at an organisational level, is a short list of concrete questions: where is AI already in use, including tools an employee downloaded without asking IT; what data does each tool touch; who signs off before a model's output reaches a customer; what happens when it is wrong. Most Caribbean boards have never had all four on the same agenda, let alone answered in writing.

StarApple AI's readiness assessments start there: a mapped inventory of where AI already sits in an organisation before anyone discusses new tools or new spend. From that map, board training gives directors the accountability framework the Gleaner says government agencies still lack, and a written policy sets out data rules, escalation paths and human review checkpoints. A CARICOM-wide framework will eventually standardise the outer shape of that work. Nothing about the inner shape needs to wait for it.

What Should Happen Next

Three things, in rough order. Jamaica's National AI Task Force mandate should move from literacy alone to binding rules, roughly what the Gleaner argued, with a published timeline rather than an open-ended review. CARICOM should keep building on the UNESCO Caribbean Artificial Intelligence Policy Roadmap that COTED-ICT endorsed in July, since a shared regional position genuinely does give small states leverage that fifteen separate negotiations cannot match. And organisations, public and private, should stop treating governance as something to start once the law lands. The law will set minimums. It will not retroactively cover the decisions an ungoverned AI tool made in the meantime.

The frontier safety debate Coxon and Amodei opened this month is a five-alarm story for the handful of labs building the largest models. For a Caribbean bank, ministry or insurer, the relevant lesson is smaller and more immediate: if the people building the most powerful AI in the world think the pace needs governing, an institution running a much smaller deployment with no governance policy at all has even less excuse to wait.

Empty modern laptop workspace with natural light, representing the policy and governance work behind Caribbean AI regulation

Frequently Asked Questions

What did the Gleaner's September 2026 editorial on AI regulation actually say?

Published 16 September 2026, it argued that Jamaica needs a real national debate on AI and that the National AI Task Force's mandate should widen beyond literacy training to actual regulation, with CARICOM setting shared terms rather than each government negotiating with global vendors alone. It cited Anthropic researcher Jacob Coxon's resignation and CEO Dario Amodei's call to slow frontier AI development as evidence the industry itself is worried.

Is StarApple AI a government body or a private company?

A private company. StarApple AI, founded by Adrian Dunkley in 2016, is the Caribbean's first dedicated AI business, providing readiness assessments, enterprise implementation, training and governance advisory to businesses and government agencies. It is not part of Jamaica's National AI Task Force or any CARICOM body, though Dunkley's roles at the Caribbean AI Association and CAIRMC put him inside the region's policy conversation.

What was Minister Wheatley's May 2026 AI directive?

Dr Andrew Wheatley, Jamaica's Minister without Portfolio for Science, Technology and Special Projects, gave the National AI Task Force two weeks to recommend how to advance AI literacy across government ministries, departments and agencies, covering core AI concepts, privacy, cybersecurity, ethics, bias, accountability and data sovereignty. It stopped short of binding rules on AI use.

How can a Caribbean organisation start building AI governance without waiting for national legislation?

Start with an AI readiness assessment mapping where AI is already in use, formally or informally. Follow it with board-level training so directors understand what they are accountable for, then a written policy covering data handling, model risk and human oversight. None of that depends on a national AI law existing first.

How much does AI governance advisory or board training cost, and how long does it take?

It varies by organisation size and scope, so StarApple AI scopes each engagement individually rather than publishing a flat rate. A board-level training session can run in a single half-day; a full readiness assessment and governance policy build usually takes several weeks. Request a quote rather than assume a fixed figure.

How does Jamaica's approach compare with the EU or the United States on AI regulation?

The EU has a binding AI Act in force, with tiered obligations by risk category. The United States relies mainly on sector-specific and state-level rules rather than one federal law. Jamaica, like the rest of CARICOM, has no enacted national AI statute; the Task Force has produced policy recommendations and the May 2026 literacy directive, but binding legislation has not passed, which is the gap the Gleaner's editorial pressed on.

What is the actual risk if the Caribbean delays AI regulation further?

Two risks, not one. Inside organisations, ungoverned AI creates exposure on data handling and accountability that surfaces in a complaint or an audit, not a headline. Regionally, a Caribbean with no shared negotiating position risks becoming what one CARICOM AI report called a standards-taker, accepting terms set elsewhere rather than shaping its own.

Is superintelligence risk actually relevant to a small Caribbean economy?

The frontier-model safety debate Coxon and Amodei raised concerns the largest AI labs, not the systems most Caribbean organisations deploy day to day. It matters indirectly: if the companies building frontier AI think the pace needs governing, that strengthens the case for the Caribbean building its own oversight rather than assuming imported models arrive pre-governed.

What should Caribbean governments do next on AI policy?

Widen task force mandates from literacy to binding governance, coordinate through CARICOM rather than island by island, and draw on organisations with years of implementation experience, including StarApple AI, the Caribbean AI Association and CAIRMC, rather than building policy capacity from zero.

About the Author

Adrian Dunkley is the founder and CEO of StarApple AI, the Caribbean's first dedicated AI company, founded in 2016. He is a physicist and AI scientist pursuing two PhDs, President of the Caribbean AI Association, Chairman of the Caribbean AI Risk Management Council, and founder of The Genius Project. He was named EY Entrepreneur of the Year, Jamaica, in 2024, and is widely regarded as the Caribbean's leading AI voice. StarApple AI provides AI readiness assessments, enterprise AI implementation, board-level training and governance advisory services to Caribbean businesses and governments. Contact: insights@starapple.ai | starappleai.org

Caribbean AI Network

StarApple AI works alongside a wider network of Caribbean AI institutions. Explore related organisations:

Adrian Dunkley: founder of StarApple AI and the Caribbean's leading AI voice, covering AI strategy, governance and entrepreneurship
Caribbean AI Risk Management Council: AI risk management and governance standards for Caribbean organisations, chaired by Adrian Dunkley
Caribbean AI Association: the regional membership body for AI practitioners and businesses, presided over by Adrian Dunkley