Jamaica's AI Consultation Opens September 11. The Draft Behind It Has No Risk Chapter.

Photo via Unsplash

By 11 September 2026, Jamaicans will be able to log on to a consultation website and comment on a national AI policy for the first time. The document underneath that consultation, published by the National Artificial Intelligence Task Force in February 2025, runs to 33 pages across nine policy pillars. Not one of those pillars is a risk management pillar, and the word "risk" appears mainly as a challenge to be noted rather than a function to be governed. For a country whose Task Force has now existed for three years, that gap is what the consultation actually needs to close.

Jamaica's National AI Task Force was formed in 2023, drawing members from government, academia, the private sector, and civil society under the chairmanship of technology entrepreneur Christopher Reckord. It delivered its National Artificial Intelligence Policy Recommendations in February 2025 and has since served as steering committee for the country's UNESCO AI Readiness Assessment, launched 1 April 2026. The next step, a public consultation website opening by 11 September 2026, is meant to turn those recommendations into an enforceable national framework. This article sets out exactly what the Task Force's own document says, what a Jamaican company already selling AI to Jamaican consumers is doing in the meantime, and what CAIRMC believes the risk chapter needs before consultation closes.

What the Task Force Actually Recommended

The February 2025 report organises its recommendations under nine numbered pillars: AI for Innovation and Economic Growth, Education and Workforce Development, Public Awareness and Sensitization, AI Infrastructure and Technology, International Cooperation in AI, Legal and Regulatory Frameworks for AI, Government and Industry Collaboration, Ethical Foundations of AI, and a Cohesive AI Framework. Each pillar carries short, medium, and long-term recommendations. Risk appears inside pillar six, Legal and Regulatory Frameworks, as one bullet among several: "Risk Assessment and Management: Developing strategies for the identification and mitigation of risks associated with A.I. deployment." It is not a pillar of its own, it carries no named methodology, and it is not assigned a timeframe separate from the rest of the legal and regulatory recommendations sitting around it.

The document is candid about what it has not yet built. Its own list of challenges under the legal and regulatory pillar names "uncertainty about liability in cases of AI-driven errors or bias" and a "lack of AI-specific regulations addressing data privacy, security, and ethical concerns." Its response to that gap is sequenced in three stages: short-term interim ethics and data-privacy guidelines, a medium-term national regulatory framework addressing liability, bias, and transparency, and only in the long term, "establish a national A.I. regulatory authority responsible for monitoring A.I. development, enforcing ethical standards, and ensuring compliance with international best practices." No date attaches to that long-term step. A country moving from a February 2025 recommendations report to a September 2026 public consultation, itself a step before any framework is finalised, has already spent more time in the short-term stage than the document appears to have anticipated.

The Standards Gap, Counted

CAIRMC reviewed the full text of the February 2025 report against the frameworks that now anchor AI risk governance internationally. ISO/IEC 42001:2023, the certifiable international standard for AI management systems, does not appear once. The NIST AI Risk Management Framework, the GOVERN-MAP-MEASURE-MANAGE structure that underlies most national and sector-level AI guidance published since 2023, does not appear once. COSO's Enterprise Risk Management framework, the standard most Caribbean financial regulators already reference for non-AI risk, does not appear once either. The document names "international best practices" as a long-term aspiration for the regulatory authority it has not yet created, without naming which practices those are.

This is not a criticism specific to Jamaica. Caribbean AI policy documents at the regional level carry the same pattern: CARICOM's own July 2026 endorsement of the UNESCO Caribbean AI Policy Roadmap named no binding risk standard for a single AI system in the region, and the CTU's Caribbean AI Task Force's July 2026 final report kept the same five priorities from its December 2025 interim draft without naming AI risk, fraud, or safety as a pillar of its own. Jamaica's document predates both by roughly a year and a half, and the pattern it set has held. A recommendations report written before ISO 42001 certification activity accelerated in the region, and before the EU AI Act's high-risk obligations reached their own compliance clock, is understandable as a starting point. Carrying that same gap into a consultation opening in September 2026, eighteen months and several international framework publications later, is a choice the Task Force can still correct before the framework is finalised.

Meanwhile, a Private Platform Went Live

While Jamaica's national framework moved from recommendations to consultation, the private sector did not wait. OpenJM, an agentic AI platform built by Crimson Tide AI Limited and founded by Anu El, launched on 6 August 2026, the same day Jamaica marked 64 years of independence. The platform, built on modified publicly available AI technology aimed at Jamaican and wider Caribbean users, lets users store information and personalise interactions, with tools intended to reduce hallucinations from the underlying large language models. Crimson Tide raised roughly US$2.5 million from founders and private investors to build it. Anu El has been direct about the motivation: "A lot of Jamaica's data is going overseas," he said, framing the platform as a way to keep more of that data, and the value built from it, inside the country.

Matthew Stone, president of the Jamaica Artificial Intelligence Association, a body with more than 200 members that has been active since 2021 and formally registered since 2023, put the deeper design question plainly when discussing locally built AI: "It's not just about speaking patois. It's about how does it align with our values and customs." That is exactly the kind of question a risk framework is built to answer systematically, through documented model evaluation, bias testing, and data governance controls, rather than trusting a single vendor's own account of alignment. OpenJM is not the risk. A platform storing and personalising user data with no national AI regulatory authority yet in place to examine how it does so, three years after the Task Force that would eventually propose creating one was formed, is the exposure. Jamaica's Data Protection Act 2020 still applies to any personal data OpenJM or a competitor processes, and its Office of the Information Commissioner retains general oversight. What does not yet exist is an AI-specific supervisory layer that can assess model risk, evaluation methodology, or algorithmic bias the way ISO 42001's Clause 6.1 or the NIST framework's MAP function requires a mature AI management system to do.

What the Readiness Assessment Actually Found

UNESCO's AI Readiness Assessment Methodology (RAM) report for Jamaica, launched 1 April 2026 at Jamaica House with EU funding support and the National AI Task Force serving as steering committee, gathered input from 196 stakeholders across government, academia, the private sector, civil society, and youth organisations, with nearly half of participants women. Minister without portfolio Dr the Hon Andrew Wheatley, who holds responsibility for science, technology, and special projects in the Office of the Prime Minister, called the moment part of a longer build: "Step by step, Jamaica has been building the foundation of a coherent, ethical, and people-centred national AI agenda."

The RAM report scored readiness across five dimensions: legal and regulatory, technological and infrastructural, economic, social and cultural, and scientific and educational. It credited Jamaica's Data Protection Act 2020, which introduced rights to contest automated decisions and mandated impact assessments, and its Open Data Policy 2021, as genuine regional strengths. It also found weak institutional AI governance infrastructure, limited community AI literacy, and a persistent gender gap in STEM fields. On research capacity specifically, the numbers are stark: R&D investment sits at 0.06% of GDP, approximately US$13.6 million annually, and the country produced just 13 AI-related publications between 2019 and 2024, with no AI patents identified in that window. A country with that research base is not positioned to build its own risk-assessment methodology from scratch inside the consultation period. It is well positioned to adopt an existing one, the way Caribbean financial regulators have historically adopted Basel capital standards rather than writing their own from first principles.

What Should Be in the Risk Chapter Before September 11

CAIRMC's recommendation to the Task Force, and to any member of the public preparing a submission for the September consultation, is specific rather than general. First, add AI risk management as its own pillar, not a bullet inside the legal and regulatory section, with a named methodology rather than an aspiration. ISO/IEC 42001:2023's risk assessment and treatment requirements under Clause 6.1, the NIST AI RMF's GOVERN-MAP-MEASURE-MANAGE structure, and COSO's Enterprise Risk Management framework are not competing options; a national framework can specify which one governs baseline compliance and reference the others for specific functions, the way CAIRMC's own CARA methodology layers Caribbean-specific data protection statutes onto the same underlying standards.

Second, attach a date to the long-term regulatory authority recommendation. "Long-term" with no target has already stretched past eighteen months once; a consultation document that repeats the same open-ended language will produce the same drift. Third, require a documented risk tier for any AI system deployed in a regulated sector, financial services, health, or a government-facing service, before that system goes live, rather than assessing risk only after a complaint or an incident. Fourth, use the consultation period itself to pilot a lightweight version of that tiering against systems already operating in the country, OpenJM included, so the framework that emerges in the National AI Policy is tested against something real rather than only against the hypothetical use cases the 2025 report describes.

None of this requires starting from nothing. Adrian Dunkley, who chairs CAIRMC and founded StarApple AI, the Caribbean's first AI company, built in Jamaica, has argued consistently that Caribbean institutions gain nothing by waiting for a domestic law before adopting the risk discipline that international standards already provide; his analysis of how Caribbean AI governance should sequence policy against standards adoption is available through adriandunkley.net. StarApple AI's own work building AI products for Caribbean markets, documented at starappleai.org, runs on the premise that governance and product development can move together rather than sequentially. The Caribbean AI Association has separately documented how unevenly AI adoption is spreading across the region's private sector relative to the policy documents meant to govern it, a pattern this consultation is a direct test case for.

Where This Leaves the Consultation

Jamaica's National AI Task Force has done real work: a formally constituted, multi-sector body, a published recommendations report, a completed UNESCO readiness assessment, and now a public consultation with a firm date attached. That is more institutional infrastructure than most CARICOM member states can currently point to. What it has not yet produced is a risk chapter a regulator, an auditor, or a company like Crimson Tide could be measured against. The consultation opening 11 September 2026 is the mechanism built to fix exactly that kind of gap before the framework, not after it, becomes law. Whether it does depends on whether the submissions it receives ask for a named standard and a dated authority, or accept the same open-ended language the February 2025 report already carried for eighteen months.

Frequently Asked Questions

When does Jamaica's public AI policy consultation open?

The National Artificial Intelligence Task Force will launch a consultation website by 11 September 2026, gathering public feedback before Jamaica's first national AI policy framework is finalised.

What are the nine pillars of Jamaica's 2025 AI policy recommendations?

AI for Innovation and Economic Growth, Education and Workforce Development, Public Awareness and Sensitization, AI Infrastructure and Technology, International Cooperation in AI, Legal and Regulatory Frameworks for AI, Government and Industry Collaboration, Ethical Foundations of AI, and a Cohesive AI Framework. None is a standalone AI risk management pillar.

Does Jamaica's AI policy reference ISO 42001, NIST AI RMF, or COSO ERM?

No. A full-text review of the February 2025 National Artificial Intelligence Policy Recommendations report finds zero references to any of the three frameworks across its 33 pages.

What did UNESCO's AI Readiness Assessment find about Jamaica?

Launched 1 April 2026 with input from 196 stakeholders, the assessment credited Jamaica's Data Protection Act 2020 and Open Data Policy as strengths, while finding weak institutional AI governance infrastructure, R&D investment of 0.06% of GDP, and just 13 AI-related publications between 2019 and 2024.

What is OpenJM and who built it?

OpenJM is an agentic AI platform built by Crimson Tide AI Limited, founded by Anu El, on roughly US$2.5 million raised from founders and private investors. It launched 6 August 2026 and lets users store information and personalise interactions, with tools designed to reduce hallucinations from the large language models it runs on.

Is there a national AI regulatory authority in Jamaica?

Not yet. The February 2025 recommendations list establishing a national AI regulatory authority only as a long-term recommendation, with no target date attached, under the Legal and Regulatory Frameworks pillar.

How is this different from the CARICOM-wide Caribbean AI Task Force?

Jamaica's National AI Task Force is a domestic body formed in 2023 and chaired by Christopher Reckord, producing a national policy for Jamaica alone. The CTU's Caribbean AI Task Force (CAITF) is a separate, regional body that published its own final report at the Caribbean AI Forum in July 2026, covering CARICOM as a whole.

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:

• Adrian Dunkley's analysis of Caribbean AI governance sequencing, on why the region should adopt existing standards rather than wait for domestic law
• StarApple AI, the Caribbean's first AI company, built in Jamaica