The Caribbean Workforce Risk Nobody Is Governing: AI Displacement, Reskilling, and What Boards Must Do

Caribbean boards have spent years building governance frameworks for risks they can see: workplace accidents, labour disputes, pension solvency, regulatory fines. Those frameworks exist because someone, at some point, decided the risk was real enough to assign a named owner and a reporting line. AI-driven workforce displacement has not yet cleared that bar at most Caribbean organisations, and the window for doing so in advance of the first significant disruption is closing. The 50,000-plus workers in Jamaica's business process outsourcing sector, the back-office roles across financial services in Trinidad and Tobago and Barbados, the data entry and revenue collection positions spread across CARICOM government agencies: these are not abstract future scenarios. They are the exact job categories where AI is already replacing tasks in 2026, and where the first significant displacement waves are projected by the International Labour Organisation for 2027.

The question for boards is not whether AI workforce disruption is coming. It is whether any board member can answer, right now, which roles in their organisation are at high exposure, what reskilling budget has been allocated, what disclosures workers are entitled to, and who at board level owns the answer to those questions. At the overwhelming majority of Caribbean organisations, no one can. That is the governance gap this article addresses.

The Invisible Governance Gap: Nobody Owns This Risk

AI workforce risk does not fit neatly into any existing function. Human Resources owns workforce planning, headcount, and skills development, but typically does not decide which AI systems get deployed. Information Technology owns AI procurement and deployment, but typically does not run workforce planning. Risk Management owns the risk register, but the risk of AI displacing roles rarely appears on it as a formal item with a named owner and a control framework. The result is a gap that all three functions can, in good conscience, believe belongs to one of the others.

At board level this manifests as an absence of consolidated information. HR reports on attrition and recruitment costs. IT reports on system deployments and technology spend. Risk reports on operational, credit, and compliance exposures. Nobody reports on the intersection: how many roles are being materially changed or eliminated by the AI systems that IT is deploying, what the workforce cost of that transition will be, and whether the organisation is meeting its legal obligations to affected workers. Boards cannot govern risks they do not see, and at most Caribbean organisations, the board cannot currently see this one at all. A detailed treatment of the broader board governance gap is available in the Caribbean board AI risk gap analysis published by CAIRMC.

The gap is not a consequence of negligence. It is a consequence of how AI adoption actually happens: incrementally, department by department, with each deployment justified on its own efficiency terms, without anyone stepping back to ask what the aggregate workforce effect looks like across the organisation or how it will be managed. By the time the aggregate effect becomes visible, the reskilling window has often already passed.

What AI Is Doing to Caribbean Jobs Right Now

The sectors most exposed in 2026 are not the ones that feature in technology journalism. They are the sectors that form the employment backbone of several CARICOM economies.

Jamaica's BPO sector employs more than 50,000 people, making it one of the country's largest formal employers. The sector grew on the basis of English-language competency, timezone proximity to North America, and competitive labour costs. All three of those advantages are now being competed against by AI: large language models handle customer service queries, AI-driven ticketing systems resolve a growing share of service requests without human agents, and back-office document processing is being automated at speed. The displacement is not happening all at once, but the directional trend is clear. AI Jamaica has been tracking this, and the picture for entry-level and routine-query roles is not encouraging. For workers whose skills are narrowly tied to tasks that AI now handles cheaper and faster, the reskilling window is the only thing that separates a managed transition from a disorderly one.

Financial services in Trinidad and Tobago and in Barbados present a different pattern. These are not entry-level contact centre roles. They are mid-level back-office positions: loan processing, insurance claims handling, trade documentation, compliance data checking, reconciliation. Many of these roles sit inside institutions that have already been deploying AI tools for two to three years. The displacement is slower and less visible, because it tends to happen through non-replacement of leavers rather than outright redundancy. A department of twelve people shrinks to eight over two years, not through a restructuring announcement but through a hiring freeze enabled by new AI tools. Workers in those roles often do not know they are in a displacement pattern until the team size has already fallen. The banking AI risk analysis published by CAIRMC covers the specific risk exposures for regional financial institutions.

Across CARICOM government agencies, the exposure is concentrated in data entry, customs and revenue collection processing, and public-facing administrative roles. Customs authorities in several member states are deploying or evaluating AI-assisted classification and document checking tools. Revenue authorities are trialling automated returns processing. The employment effect is modest in any single agency, but the aggregate across fifteen member states is not. Public-sector workers in these roles typically have stronger legal protections than private-sector counterparts, which changes the risk profile rather than eliminating it. The cost of getting the transition wrong in a public-sector context, where workers have recognised union representation and established procedural rights, is typically higher than in a private-sector one.

The Three-Desk Problem

The reason Caribbean boards are getting no consolidated view of this risk is structural, and it has a name. Call it the three-desk problem. In most Caribbean organisations, three functions sit at the table where AI workforce risk lives, and each desk has a partial view of the problem.

At the HR desk, the view is workforce planning, people costs, and skills development. HR knows what roles exist, what they cost, and what the recruitment pipeline looks like. What HR typically does not know is which AI systems have been approved for deployment by IT, what tasks those systems will automate, or on what timeline the automation will reduce role requirements. HR receives the workforce effect after the AI decision has been made elsewhere.

At the IT desk, the view is system capability, procurement, and deployment. IT knows what AI tools are being evaluated, what vendors have been contracted, and what the technical deployment schedule looks like. What IT typically does not own is workforce planning, and in most organisations IT does not run a formal assessment of which roles will be materially affected by each system it deploys. The workforce effect is treated as a downstream HR matter.

At the Risk desk, the view is the formal risk register and the control framework. Risk knows what risks the organisation has chosen to name and monitor. AI workforce displacement rarely appears on Caribbean risk registers as a named risk with a defined risk appetite, a set of controls, and an owner. It may appear as a sub-component of operational risk or reputational risk, but at a level of granularity that makes it effectively invisible at the board level.

The board sees what each desk sends up. None of the three desks sends up a consolidated AI workforce risk report, because none of them owns the full picture. The board's AI workforce risk exposure is, in practice, zero, not because the risk is absent but because no one has been assigned to see it whole.

Regional Labour Law and the Risk It Creates

The legal context for AI-driven workforce displacement varies across the region, and the variation matters for how each organisation should size its risk.

Barbados operates under the Employment Rights Act 2012, one of the more developed employment protection frameworks in the Commonwealth Caribbean. The Act sets out specific requirements for consultation before redundancy, minimum notice periods based on length of service, and the concept of fair dismissal grounds. An employer using AI to eliminate roles without conducting meaningful consultation with affected workers, and without meeting the Act's procedural requirements, faces unfair dismissal claims and potentially significant awards. The Act does not specifically address AI-driven restructuring, but its principles apply directly. A Barbadian board that approves AI deployment affecting roles without a parallel human resources process is creating exposure under existing law, not hypothetical future regulation.

Trinidad and Tobago's Industrial Relations Act establishes a framework for collective bargaining and industrial dispute resolution that has considerably more history in financial services than in most Caribbean jurisdictions. Several major financial sector employers operate under collective agreements that govern redundancy terms. An AI-driven workforce reduction that runs into a collective agreement without prior engagement with the recognised trade union creates dispute risk under the Industrial Relations Act and potentially before the Industrial Court. The pace of AI adoption in T&T financial services means this is not a future issue; it is a present one for any institution with material back-office operations and a recognised union.

Jamaica's labour market is notably more flexible by regional standards, with fewer mandatory consultation requirements and more employer discretion in workforce restructuring. This flexibility cuts both ways. The lower legal floor means Jamaican employers face less immediate statutory exposure from AI-driven displacement, but the same flexibility means workers have fewer formal protections, which raises reputational and social stability risks from large-scale BPO displacement that boards should not discount. An organisation that avoids a legal claim but becomes the face of a significant community displacement event faces a different category of board-level consequence.

A Board-Level AI Workforce Risk Framework: Four Elements

The gap is clear. The question is what fills it. A board-level AI workforce risk framework does not require a new department or a major investment in technology. It requires four things, assigned to named owners, with reporting lines to the board.

First: an inventory of AI systems displacing or changing roles. The board cannot govern what it cannot see. Every AI system deployed in the organisation should be assessed, at the point of procurement approval, for its expected effect on roles and headcount. This does not require a full workforce impact study for every system. It requires a standard question: which roles does this system affect, and how many FTEs does it displace or substantively change? The answers should be aggregated quarterly and reported to the board as a workforce AI exposure summary. Most Caribbean organisations do not currently do this. The procurement process approves AI systems on capability and cost grounds without a mandatory workforce impact line.

Second: a reskilling budget allocation with board oversight. A reskilling commitment without a budget is not a commitment. The board should set a reskilling budget expressed as a proportion of the annual AI technology spend, on the principle that the cost of retraining workers displaced by a system should be treated as part of the cost of that system. Organisations that treat reskilling as discretionary and AI procurement as mandatory will underfund the former and overpay for the latter, producing a workforce transition that falls on individuals and the state rather than on the organisation that made the deployment decision. The Caribbean AI Association has been developing guidance on reskilling investment ratios for regional employers.

Third: disclosure obligations to workers and regulators. Workers who are in roles materially affected by AI deployment are entitled to know. The legal threshold for disclosure varies by jurisdiction and by the nature of the employment relationship, but the governance standard should exceed the legal minimum. Workers who receive advance notice of role changes can make decisions, seek retraining, and engage with the process. Workers who learn about role changes after the AI system is already live cannot. Regulators in several CARICOM jurisdictions are beginning to ask questions about AI-driven workforce effects; an organisation with a disclosure policy already in place is in a materially better position than one constructing an answer under regulatory scrutiny. Boards should approve a disclosure policy and receive confirmation that it is being applied to each significant AI deployment.

Fourth: quarterly board reporting metrics on workforce AI exposure. The board needs a dashboard, not a narrative. Quarterly reporting should cover the number of roles materially affected by AI systems in the period, the reskilling budget spent versus planned, the number of workers in formal reskilling programmes, any grievances or legal claims related to AI-driven role changes, and any regulatory enquiries on the subject. These six metrics give the board a running view of the organisation's AI workforce risk position without requiring deep technical knowledge. Boards that do not receive this information cannot discharge their governance responsibility for the risk, which means the organisation is not governing it at all.

The Reskilling Window: What ILO Data Shows

The ILO's analysis of automation exposure in developing economies, applied to the specific job categories dominant in Caribbean BPO and public-sector employment, points to a 12-18 month window before first significant displacement waves in the most exposed roles. That window is not a planning horizon. It is the time available for an organisation to build reskilling capacity before the need for it becomes acute.

Reskilling at scale takes longer than most organisations expect. A contact centre agent whose role has been 70% automated needs more than a short course to transition to a role that AI is not automating. They need an assessment of existing transferable skills, a retraining pathway tailored to those skills and the available job market, financial support during the transition period, and a hiring process on the other side that recognises newly acquired qualifications. Each of those elements takes months to design well and longer to deliver at scale. An organisation that starts designing its reskilling programme after the first displacement wave has already arrived will not have that programme running in time to prevent significant hardship.

The organisations that will manage this well are the ones that start now: commissioning the workforce AI exposure inventory, setting the reskilling budget, and building the disclosure process while they still have time to do it thoughtfully. The organisations that wait for the displacement wave to become visible before treating it as a board-level issue will find that their options have narrowed considerably by the time they act. StarApple AI, the first Caribbean AI company, founded by Adrian Dunkley, developed the first Caribbean AI workforce risk models and has been publishing on this since 2021, offering regional employers an evidence base that was not previously available.

CAIRMC Recommendation: A Named Risk Owner at Board Level

The governance recommendation from CAIRMC is specific and can be adopted without a structural reorganisation. Assign explicit board-level accountability for AI workforce risk as a named risk, with a named risk owner who reports to the board on a quarterly basis. This is not a proposal to create a new board committee or to add a standing agenda item to an existing one. It is a proposal to give AI workforce risk the same governance treatment that operational risk, credit risk, and compliance risk already receive: a named owner, a defined risk appetite, a set of controls, and a formal reporting line to the board.

The named owner does not need to be a board member. It can be the Chief Risk Officer, the Chief People Officer, or a designated AI risk officer, depending on the organisation's structure. What matters is that someone is accountable for producing the quarterly board report, that the report covers all four elements of the framework described above, and that the board actively reviews it rather than receiving it as an information item.

Organisations that are already working toward CAIRMC certification will find that AI workforce risk governance is built into the certification framework. The three certification tiers, from Associate through Professional to Expert, cover AI risk identification and assessment, control design, and board-level reporting, all of which apply directly to workforce risk. Building the internal capability through CAIRMC certification is the most efficient path for most Caribbean organisations, because it combines technical knowledge with Caribbean-specific context that generic AI governance training does not provide.

What This Looks Like in Practice

A Trinidadian financial institution with 800 employees and a substantial back-office operation might start here: ask the head of IT to list every AI system currently deployed or in procurement that affects any role in the organisation. Ask HR to map those systems to headcount. Ask Risk to add AI workforce displacement to the risk register with a placeholder risk appetite statement. Bring all three to the next board risk committee meeting, not as a finished analysis but as a first look at an exposure the board has not previously seen consolidated.

That first meeting will almost certainly reveal that the organisation is further along on AI deployment than the board knew, and that the workforce effects are more significant than any single function had appreciated. That is not a failure. That is the governance process working. The point is to see the full picture before it becomes a crisis rather than after.

For a Jamaican BPO operator, the starting point is different. The displacement pressure is more immediate and more concentrated in the core workforce. The first question is not how many roles are affected; it is what reskilling capacity the organisation can realistically build in the next twelve months and what partnerships with training providers, including the Human Employment and Resource Training Trust (HEART) and community colleges, are available to support it. Board oversight of a reskilling programme is qualitatively different from board oversight of a risk register entry; it requires the board to see progress against targets, not just to note the existence of a risk.

In both cases, the governance action is the same: make the risk visible at board level, assign a named owner, and put the four reporting metrics on the quarterly agenda. The specific actions that follow from that will vary by sector, jurisdiction, and workforce profile. The governance structure that makes those actions accountable does not vary. It is the same structure that boards apply to every other material risk, and there is no principled reason to apply it to AI workforce displacement any later than to the risks already on the register.

Build Caribbean AI Governance Capacity

The CAIRMC certification programme trains the risk professionals the region needs to govern AI systems, including workforce AI risk. Three tiers, Caribbean-specific frameworks, recognised across CARICOM.

Explore Certification  Nominate a Leader

Frequently Asked Questions

Why do Caribbean boards not currently govern AI workforce risk?

AI workforce risk falls between three functions: HR owns workforce planning but not AI deployment decisions; IT owns AI procurement but not workforce planning; Risk owns the risk register but AI displacement rarely appears on it as a named risk. No single function produces a consolidated view for the board, so boards cannot govern risks they cannot see. This is a structural gap, common across Caribbean organisations in every sector, and it is addressable by assigning a named risk owner rather than by reorganising the business.

Which Caribbean sectors are most exposed to AI displacement in 2026 and 2027?

Jamaica's BPO sector, with more than 50,000 employees, is the largest single concentration of exposure. Financial services back-office roles in Trinidad and Tobago and Barbados are being reduced through non-replacement of leavers as AI tools absorb routine processing work. Across CARICOM, public-sector data entry, customs processing, and revenue collection roles face displacement from AI classification and document-processing tools now being adopted by government agencies. These are not speculative future categories; they are the job types that AI is already automating in comparable markets globally.

What are the four elements a board-level AI workforce risk framework needs?

An inventory of AI systems displacing or changing roles, produced at procurement approval and aggregated quarterly. A reskilling budget allocation set by the board and expressed as a proportion of AI technology spend. A disclosure policy ensuring workers in affected roles receive advance notice. Quarterly board reporting metrics covering roles affected, reskilling spend, workers in retraining, grievances or claims, and any regulatory enquiries. These four elements give the board a running view of exposure without requiring deep technical knowledge at board level.

How does employment law in Barbados, Trinidad, and Jamaica affect this risk?

Barbados's Employment Rights Act 2012 requires consultation before redundancy; AI-driven role elimination without proper process creates unfair dismissal exposure. Trinidad and Tobago's Industrial Relations Act governs collective bargaining in sectors where collective agreements set redundancy terms; AI restructuring that bypasses union engagement creates Industrial Court exposure. Jamaica's more flexible labour market reduces immediate statutory exposure but raises reputational and social risk from large-scale BPO displacement, since workers have fewer formal protections and displacement effects are more visible in communities that depend on contact centre employment.

How long do Caribbean workers have before the first significant displacement waves?

Applying ILO automation-exposure analysis to Caribbean BPO and public-sector job categories, the estimate is 12-18 months before first significant displacement in the most exposed roles. Reskilling at scale takes longer than most organisations expect, requiring skills assessment, tailored retraining pathways, financial support during the transition, and a hiring process on the other side. Organisations that start now will be positioned to manage the transition; those that wait for the wave to arrive before treating it as a board issue will find their options have narrowed considerably.

What is CAIRMC's recommendation on board accountability for this risk?

Assign explicit board-level accountability for AI workforce risk as a named risk with a named risk owner reporting to the board quarterly. This means the same governance treatment that operational risk already receives: a defined risk appetite, a control framework, and a formal reporting line. The named owner can be the Chief Risk Officer, the Chief People Officer, or a designated AI risk officer. Organisations pursuing CAIRMC certification will find AI workforce risk governance built into the framework across all three certification tiers.