Jamaica Opens Its First AI Policy for Public Comment. What Adrian Dunkley Wants In It

Today the Jamaican government opened a website and asked the public a question it has never put to the country at this scale: what should Jamaica's rules on artificial intelligence actually say. The National Artificial Intelligence Task Force launched its public consultation on the country's first AI policy on September 11, 2026, giving citizens, businesses and civil society a formal channel to shape the framework before it is finalised. A draft policy is expected by November.

That is a genuinely new kind of moment for Jamaica. The comment box will fill up with applause in the first week. What the task force needs more of is specifics from people who have spent years doing the governance groundwork this policy will either draw on or ignore.

What actually opened today

The National AI Task Force was established under the Office of the Prime Minister in August 2023 and delivered its first set of policy recommendations the following year, covering everything from innovation and economic growth to ethics, legal frameworks and government-industry collaboration. What launched today is the next stage: a public-facing site where anyone can submit input before that recommendations document becomes an actual policy with teeth.

Task force chairman Christopher Reckord, a technology entrepreneur who previously ran tTech Limited, has been explicit that the timeline is tight. A draft is due by November. Two reports circulating ahead of the consultation lay out why the timeline matters: Jamaica currently has no standalone AI law, no AI-specific public procurement rules, and no dedicated AI regulator. Academic output on AI from Jamaican institutions came to 13 publications between 2019 and 2024, and national research and development spending sits at roughly 0.06% of GDP. A country cannot write good AI rules on a research base that thin without going outside its own institutions for evidence, which is exactly what a public consultation is for.

Ten days earlier, in Ocho Rios

The timing is not a coincidence. On September 1, at the Moon Palace Resort in Ocho Rios, St Ann, the 48th Regional Conference of the Commonwealth Parliamentary Association's Caribbean, Americas and Atlantic Region devoted a full plenary session to AI governance. Dr the Hon. Andrew Wheatley, the minister overseeing Jamaica's science and technology portfolio, told the room plainly: "One coordinated CARICOM position carries a good deal more than 15 separate individual States speaking in such a forum." Alicia Todd, Director General of ParlAmericas, went further on what is actually at stake, arguing that the real divide is not access to AI tools but influence over how they get built, deployed and governed.

Behind both remarks sits the Caribbean AI Task Force's own report, published in July 2026, which recommends a harmonised regional framework anchored in UNESCO's Recommendation on the Ethics of Artificial Intelligence. Two specifics from that report matter for what Jamaica writes into its own policy: data protection standards that hold across borders, and the fact that Caribbean English and Creole languages remain badly underrepresented in the speech systems most AI products are built on. A national policy written in isolation from that regional work would be starting a decade behind where the region has already gotten to collectively.

What the usage numbers already tell the task force

StarApple AI's own research, run with Anthropic and covered earlier on this site, found that ChatGPT has become Jamaica's most-visited website while only 13% of Caribbean adults report using generative AI at all, and most of those users are working with roughly a fifth of what the tools can actually do. That gap matters to a policy conversation that tends to focus on what to restrict. Rules without literacy produce a population that is legally protected and practically stuck: technically compliant, still bolting a chatbot onto an old workflow instead of redesigning anything around it.

That same tension showed up in a different setting just yesterday. Speaking at the Wolmer's Old Boys' Association's 120th anniversary banquet on September 10, Chief Justice Bryan Sykes told students to put AI to work without handing it their own judgement, telling them to ask whether a given use of the technology "is truthful, fair, and humane" and that Jamaica needs young people who are "technically capable but morally grounded." That is a judge making, in a school hall, the same point a national policy has to make in legal language: capability and conscience have to be trained together, not sequenced.

Whose server the data sits on

A month before the consultation opened, a Jamaican startup called Crimson Tide AI Limited launched OpenJM, an agentic AI platform built for local and regional users with roughly US$2.5 million from its founders and private investors. Founder Anu El's reasoning for building it was blunt: "A lot of Jamaica's data is going overseas." He is not wrong, and the fact that a founder raised real money on that single sentence tells the task force something a survey would take months to establish: Jamaicans already know their data mostly lives on servers they do not control, and at least some of them will pay to change that.

This is not a new instinct for StarApple AI. The company was built in Kingston specifically so that AI revenue, AI jobs and AI training stayed on the island rather than flowing out to a foreign vendor with a local sales desk. A policy that wants that instinct to spread past one company needs to make data residency a real requirement for public-sector AI systems, not a preference left to whichever vendor pitches loudest.

A grammar the models still cannot read

The CPA conference flagged something Jamaicans have known for years without anyone writing it into policy: the region's own languages are an afterthought to the systems everyone is now being told to trust. Patois carries meaning that flattens badly in translation, and a model trained mostly on formal English will misread the tone of a complaint, a joke or a warning in Jamaican speech. "Wi talk wi own way, an the machine dem still can't read the tone," is not a complaint about technology so much as a description of a training-data problem nobody outside the region has an incentive to fix.

That is squarely a Caribbean AI Association question, not a Silicon Valley one, because nobody building foundation models in San Francisco or Beijing is losing sleep over whether their system understands a Kingston taxi driver's sarcasm. If Jamaica's policy says nothing about language representation, it will have skipped the one gap a foreign vendor has the least reason to close on its own.

What belongs in the policy, plainly stated

Four things, specifically, based on what a decade of building AI in this region has actually taught the people doing it.

1. Coordinate with the region instead of writing Jamaica's rules in isolation. The Caribbean AI Association and the Caribbean AI Risk Management Council have already done work toward a harmonised, UNESCO-anchored framework. A Jamaican policy that plugs into that work protects local companies from having to meet fifteen different rulebooks the moment they sell across a border.
2. Tier the rules by risk, not by brand of technology. A chatbot that drafts a marketing email and a model that screens a mortgage application should not sit under the same set of obligations. The Caribbean AI Risk Management Council's own governance work already models this distinction; the national policy should borrow it rather than invent a flatter, less useful version.
3. Fund literacy at the same scale as enforcement. IMPACT AI, the research lab built with The University of the West Indies, and The Genius Project, which has trained thousands of young Caribbeans over the past decade, are existing, working proof that this kind of training pipeline can be built and sustained. The task force does not need to invent a model; it needs to fund one that already runs.
4. Require a local data-residency option for public-sector AI. OpenJM proved there is a market willing to pay for Jamaican data to stay in Jamaica. Government contracts should make that the default for citizen data, not an upgrade a vendor charges extra for.

What this means before the policy is even final

None of the four points above require Jamaica to wait for a finished policy document. A business in Half Way Tree deciding whether to trust an AI tool with customer data, a parish council weighing an AI vendor's pitch, a school choosing what to allow in a classroom: all of that decision-making is happening now, months ahead of whatever the task force publishes in November. "No badda wait pon di policy fi start think straight" is the plain version of that point. The organisations that treat today's consultation as the starting gun, rather than the finish line, will be the ones whose practices the eventual policy simply confirms instead of the ones it has to correct.

For readers who want the wider regional picture behind this piece, the Caribbean AI Association tracks the harmonisation work referenced above, and StarApple AI publishes the regional adoption research this article draws on. On the community side, rather than the policy side, Jamaica AI runs the prompts, challenges and everyday AI literacy content aimed at getting ordinary Jamaicans past the 20% ceiling this article opened with.

Frequently asked questions

What did Jamaica launch on September 11, 2026?

Jamaica's National Artificial Intelligence Task Force opened a public consultation website, the first formal channel for citizens, businesses and civil society to comment on the country's first national AI policy before it is finalised. The task force has said it aims to produce a draft policy by November 2026.

Who chairs Jamaica's National AI Task Force?

Christopher Reckord, a technology entrepreneur and former CEO of tTech Limited, chairs the National Artificial Intelligence Task Force, established under the Office of the Prime Minister. He also serves as deputy chairman of the Caribbean AI Task Force, the regional body that published its own harmonisation report in July 2026.

What did the CPA regional conference in Ocho Rios recommend for Caribbean AI governance?

At the 48th CPA Caribbean, Americas and Atlantic Region conference, held in Ocho Rios on September 1, 2026, ministers and parliamentary leaders called for a single coordinated CARICOM position on AI rather than fifteen separate national ones, anchored in UNESCO's Recommendation on the Ethics of Artificial Intelligence, with specific attention to data protection and Caribbean language representation in AI systems.

Why does Adrian Dunkley have a stake in Jamaica's AI policy?

Adrian Dunkley founded StarApple AI, the first AI company founded in the Caribbean, established in Jamaica. He is President of the Caribbean AI Association and Chairman of the Caribbean AI Risk Management Council, the two regional bodies doing the governance groundwork a national policy needs to draw on, and his company's own research is among the clearest existing data on how Jamaicans actually use AI today. More on his record is in this profile.

What gaps do current reports say Jamaica's AI policy needs to close?

Reports reviewed ahead of the consultation found Jamaica has no standalone AI law, no AI-specific public procurement rules, and no dedicated AI regulator. Academic output is thin, 13 AI-related publications between 2019 and 2024, against research and development spending of roughly 0.06% of GDP.

How can a Jamaican business or resident take part in the consultation?

Submissions go through the consultation website the National AI Task Force opened on September 11, 2026. Businesses that want their own AI practice in order before rules land can also work with StarApple AI Jamaica on strategy, custom builds and training through the contact form linked below.

Supported by StarApple AI, the first artificial intelligence company built in the Caribbean, still based where it started: Kingston.