Two numbers landed on the desk of Jamaica's Ministry of Education in the same week this August, and neither one is good. The first: 4,048 registered CSEC candidates never showed up for the May and June 2026 sitting, 1,875 of them from public schools, after the government had already paid their subsidised exam fees. The second: the Caribbean Examinations Council recorded 128 confirmed exam irregularity cases at that same sitting, up from 80 the year before, with a category examiners had never had to track before now sitting alongside the usual phone-in-the-sock offences. Artificial intelligence misuse.
AI can address both crises at once: predictive absenteeism models can flag at-risk candidates weeks before a subsidised seat goes empty, and AI-calibrated integrity tools can catch undisclosed AI use in an exam hall without falsely accusing the honest majority of students who never touched a chatbot. Neither fix requires CXC to rebuild the exam system from scratch. Both require the region to stop treating AI as only a threat to exam integrity and start using it as the tool that protects that integrity.
That framing matters because the two stories are usually told separately. One is an accountability story about wasted subsidy money and parents who cannot be bothered to send their children to sit a paid-for exam. The other is a technology scare story about students using ChatGPT to cheat. Read together, against the same data set from the same exam sitting, they describe a single system under strain: a regional examination body trying to serve sixteen territories with wildly uneven resources, using processes designed for a world before generative AI existed on every phone in every exam candidate's pocket.
The Bill Nobody Wants to Pay
Jamaica subsidises exam fees in a handful of core CSEC subjects, including English Language and Mathematics, on the theory that no student should be priced out of the two qualifications that most determine access to further education and employment. In 2026, that subsidy covered 20,459 English Language candidates at an average of J$4,947.50 each, and 18,309 Mathematics candidates at the same average rate, a combined outlay of roughly J$191.8 million. The exact per-student figure varied by circumstance: candidates in parishes affected by Hurricane Melissa received J$4,690, while candidates elsewhere received J$5,205.
Dr. Kasan Troupe, Permanent Secretary in the Ministry of Education, put the accountability problem plainly in a Jamaica Gleaner report published August 30: every unjustified absence from an examination is a loss of taxpayers' money. Linvern Wright, principal of William Knibb Memorial High School, went further, calling the absences a symptom of students not taking their education seriously. Stewart Jacobs, president of the National Parent-Teacher Association, placed the responsibility with parents. Ralston Smith, an educator and social scientist, has proposed reimbursement contracts that would require families to repay subsidised fees when a student skips an exam without a valid excuse.
All four are describing the same gap from different angles, and none of them are wrong. What none of them have is a way to know, before the exam hall doors close, which of the thousands of registered candidates is actually going to be a no-show. That is precisely the kind of pattern-recognition problem AI is good at, and precisely the kind of problem the region has not yet pointed AI at.
A New Kind of Cheating
While the Ministry of Education was totting up the cost of empty desks, CXC's Director of Operations, Dr. Nicole Manning, was briefing the region on a different kind of failure. The 128 irregularity cases recorded in 2026 were up sharply from 80 in 2025, with the CSEC level alone nearly doubling, from 57 cases to 106. CAPE held roughly steady, at 22 cases against 23 the year before. The top three categories, in Dr. Manning's own words, were the use of unauthorised devices, collusion, use of unauthorised material, and artificial intelligence misuse, which she described as the new one on the block, something the council had never had to deal with before.
What "AI Misuse" Actually Means Here
CXC has not published a full breakdown of what its AI misuse cases involved, but the pattern reported across regional outlets, including the Jamaica Gleaner, the Jamaica Observer, and Barbados Today, points to generative AI chatbots being used during timed, supposedly closed-book components of the exam, work that was then submitted as the candidate's own. This is a different offence from the disclosed, school-based AI use CXC formally began permitting earlier in 2026 for certain coursework assignments. One is sanctioned. The other, used inside the exam hall itself, gets a candidate's grade cancelled or the candidate disqualified outright.
The council also recorded 359 candidates granted compassionate consideration for circumstances beyond their control, 288 at CSEC and 71 at CAPE, a slight decline from 377 the year before. Separately, CXC activated its disaster recovery protocols for Jamaican candidates after Hurricane Melissa disrupted roughly half of the island's exam centres, a reminder that absenteeism, irregularities, and climate disruption are not three unrelated problems in the Caribbean exam system. They are three symptoms of the same overstretched infrastructure.
The Territory Gap Nobody Is Fixing
Layer a third data set on top of the first two and the picture gets harder to ignore. Regional CSEC performance in 2026 sat at 72.18 percent acceptable grades, against 93.52 percent for CAPE, and the gap between the strongest and weakest territories inside those averages is far wider than the headline numbers suggest. In Mathematics, fourteen territories scored between 43 and 77 percent while seven scored between 30 and 38 percent. In Social Studies, the lowest-performing territory recorded 44 percent against a highest of 100 percent. Jamaica's own Mathematics pass rate, 48.7 percent in 2026, up from 44 percent the year before, sits above the 42 percent regional average, itself a four-point improvement on 2025.
Dr. Manning flagged the disparity between territories as a concern in her own briefing, and it is easy to see why. A student sitting the same CXC Mathematics paper in a territory that scored 77 percent is being examined against a fundamentally different level of teacher access, past-paper practice, and school resourcing than a student in a territory that scored 30 percent, even though both receive an identical CXC certificate. Subsidy reform and integrity enforcement will not close that gap on their own. Closing it requires getting more instructional support to the students furthest behind, which is exactly where AI-adaptive tutoring has already proven itself at scale in other education systems.
The Core Argument
Three CXC problems reported in the same August, wasted subsidy, doubled AI cheating cases, and a widening territory gap, are usually treated as separate failures with separate fixes. They are not. All three come down to a system that only learns what is happening after the exam is over: after the seat is empty, after the AI-written answer is submitted, after the results show a territory has fallen behind. AI applied before the exam, inside the exam, and around the exam can close that lag on all three fronts at once.
What AI Actually Does for a Caribbean Exam System
None of the four applications below require CXC to redesign its exams from the ground up. Each one layers onto data the council and regional ministries already collect, and each one targets a specific failure identified in the 2026 reporting.
1. Predictive Absenteeism Modelling
A model trained on school attendance records, mock exam sittings, prior CXC registration history, and known disruption events, such as storm damage to a given parish, can score every registered candidate's likelihood of actually sitting the exam weeks in advance, not on exam morning. A school flagged with a cluster of high-risk candidates in English or Mathematics gets a concrete list of names to follow up on: a call home, a check on transport arrangements, a conversation with a guidance counsellor. This does not replace parental responsibility, which Stewart Jacobs is right to raise. It gives schools and ministries an actual chance to act on that responsibility before J$191.8 million in subsidy is already spent on empty desks.
2. Region-Calibrated AI Integrity Detection
Generic, US-built AI detection tools are a poor fit for Caribbean students, whose English usage, sentence rhythm, and idiom differ from the American and British writing samples most detection models are trained on, and a false accusation carries the same weight as a real one when a candidate's grade is on the line. What CXC needs is a detection layer trained specifically on regional student writing samples, flagging statistically unusual answer similarity or writing-pattern shifts for human review rather than automatic disqualification. That combination, AI flags the anomaly, a trained examiner makes the final call, is the same model banks already use to catch fraud without freezing every unusual transaction automatically.
3. AI-Adaptive Tutoring for the Territories Furthest Behind
The seven territories that scored 30 to 38 percent in 2026 Mathematics are not short on capable students. They are short on consistent access to a qualified Mathematics teacher and a steady diet of exam-style practice, the two inputs that most reliably move a CSEC pass rate. Adaptive tutoring platforms that adjust question difficulty in real time based on a student's answers, already deployed at scale in parts of Sub-Saharan Africa and South Asia to raise standardised test scores in under-resourced schools, could be piloted in exactly these territories, targeted at the two subjects, Mathematics and English, where the subsidy money and the pass-rate gap both concentrate.
4. Smart Subsidy Verification
Rather than paying the full subsidy upfront for every registered candidate and discovering months later who never appeared, an AI-assisted verification step, cross-referencing registration data against school attendance and mock exam participation, could flag registrations with a genuinely low probability of resulting in an actual sitting before the subsidy is disbursed. That is not a punitive measure against students. It is a way to redirect a portion of that J$191.8 million toward the intervention support described above, catching at-risk candidates instead of simply writing off their unused subsidy after the fact.
The Stakes Beyond One Exam Sitting
CSEC and CAPE are not simply school assessments. They are the gateway credential that determines whether a Caribbean teenager can enter a regional university, apply to a UK or Canadian institution, or qualify for entry-level professional employment. A subsidy system that loses J$191.8 million to no-shows is money that did not reach the students who might have used a small transport stipend or after-school tutoring support to actually sit the exam. An integrity system that cannot reliably tell the difference between a student who used AI to cheat and a student whose honest writing happens to trip a poorly calibrated detector risks cancelling the wrong grades, in a region where a single cancelled subject can close off a university place. And a widening gap between the strongest and weakest CXC territories, left unaddressed, hardens into a regional two-tier system where a CSEC certificate from one territory quietly carries more weight than the identical certificate from another.
None of these outcomes are inevitable. All three problems were visible in the same month, from the same exam sitting, because CXC and Jamaica's Ministry of Education are already collecting and reporting the data needed to see them. What is missing is a layer of prediction and detection built to act on that data before the next sitting rather than to explain it after the fact.
What CXC and Caribbean Ministries Need to Do Now
Three steps would move this from a diagnosis to a pilot within a single exam cycle.
First, CXC and Jamaica's Ministry of Education should jointly commission a predictive absenteeism pilot ahead of the January 2027 CAPE sitting, using the 2026 attendance and registration data already on file, and measure whether flagged at-risk candidates who receive early intervention show up at a materially higher rate than those who do not. Second, CXC should fund development of a region-calibrated AI writing detector, trained on actual CSEC and CAPE answer scripts rather than adapted from a US or UK model, with a mandatory human review step built into the process before any grade is cancelled. Third, the four to six member territories with the widest Mathematics and English pass-rate gaps should be offered first access to an adaptive tutoring pilot, funded jointly through CXC's existing regional cooperation structures and each territory's own education ministry, with results tracked against the next two CSEC sittings.
An Exam System Built for the Wrong Decade
CXC has run CSEC and CAPE for decades on a model built for a slower, more analogue kind of cheating: a hidden phone, a note passed under a desk, a whispered answer. Generative AI did not create the absenteeism problem or the territory disparity, but it has exposed how much the whole system still depends on finding out what went wrong only after it has already happened. The council caught its first AI misuse cases this year because someone eventually noticed. Jamaica found out about its 4,048 no-shows on the day of the exam, not the week before.
The technology that made undisclosed AI use possible in an exam hall is the same category of technology that can flag a struggling candidate before the subsidy is spent, calibrate a fair detector instead of an imported one, and put a working tutor in front of the seven territories that need it most. The Caribbean does not need to choose between fearing AI in its exam system and using it. The 2026 sitting has already made the argument for doing both.
Frequently Asked Questions
What are CSEC and CAPE, and who runs them?
The Caribbean Secondary Education Certificate (CSEC) and the Caribbean Advanced Proficiency Examination (CAPE) are the region's main school-leaving qualifications, set and marked by the Caribbean Examinations Council (CXC), a regional body owned by sixteen English-speaking Caribbean governments. CSEC is typically sat at age 15 or 16, roughly equivalent to UK GCSEs, while CAPE is a sixth-form qualification used for university entry across the Caribbean, the UK, Canada, and the United States.
Is the CXC no-show and AI cheating problem specific to Jamaica or is it region-wide?
Both issues are region-wide, though the Jamaican figures are the most detailed publicly reported ones. More than 35,000 candidates across the Caribbean did not turn up for the May and June 2026 CSEC and CAPE sittings, and CXC's 128 recorded irregularity cases, including the first flagged instances of artificial intelligence misuse, cover candidates across all sixteen CXC member territories, not Jamaica alone.
How can AI actually reduce CXC exam no-shows?
A predictive model trained on attendance records, mock exam results, and prior no-show patterns can flag at-risk registered candidates weeks before the sitting, while there is still time for a school or the Ministry of Education to intervene with a phone call, a home visit, or transport support. Right now, the first official signal that a subsidised candidate will not appear is the empty desk on exam day, when the J$4,690 to J$5,205 per-subject subsidy has already been spent with no way to recover it.
How much money has the Caribbean lost to CXC no-shows in 2026?
Jamaica alone subsidised English Language and Mathematics CSEC entries to the tune of roughly J$191.8 million in 2026, covering 20,459 English candidates and 18,309 Mathematics candidates. Every one of the 4,048 unjustified Jamaican absences from that sitting, and every one of the more than 35,000 regional absences, represents subsidised public money spent with no educational return, according to Permanent Secretary Dr. Kasan Troupe.
What is the difference between AI exam cheating and AI exam integrity tools?
AI exam cheating is a student using a generative AI tool such as a chatbot during an assessment to produce answers that are not their own work, the practice CXC flagged for the first time in its 2026 irregularity report. AI exam integrity tools work in the opposite direction: they are systems that analyse writing patterns, flag statistically improbable answer similarity, or verify a candidate's own process through oral or practical components, giving examiners a way to detect the first kind of AI use without banning technology from the classroom altogether.
Is it risky to use AI to detect cheating on Caribbean school exams?
Yes, and the risk is real: AI detection tools built for North American or European writing samples can misread the sentence rhythms of Caribbean English and Caribbean-educated writers, producing false accusations against honest students. Any detection system CXC or a ministry deploys needs to be trained and calibrated on regional student writing, paired with a human review step before any grade is cancelled, rather than treated as an automatic verdict.
Does CXC or any Caribbean government currently regulate AI use in exams?
CXC permitted limited, disclosed AI use in some school-based assessments starting in 2026, but the irregularity cases now being flagged involve undisclosed AI use during the timed, closed-book components of CSEC and CAPE exams, which remains prohibited and can result in grade cancellation or disqualification. No CXC member territory has published a comprehensive, exam-specific AI policy that covers both permitted classroom use and prohibited exam-hall use in one framework as of August 2026.
Where is AI headed in Caribbean secondary education over the next few years?
Expect three converging trends over the next two to three years: wider CXC-sanctioned use of AI tools in coursework and school-based assessment, growing investment in AI-adaptive tutoring aimed at closing the pass-rate gap between the strongest and weakest CXC territories, and the first region-wide AI detection standards for exam-hall misuse, likely developed jointly by CXC and its sixteen member ministries of education rather than territory by territory.
Why do CSEC pass rates vary so widely between Caribbean territories?
In the 2026 Mathematics sitting, fourteen CXC territories scored between 43 and 77 percent while seven scored between 30 and 38 percent, a gap CXC's own Director of Operations, Dr. Nicole Manning, has flagged as a concern. The gap reflects uneven access to qualified subject teachers, uneven exposure to past-paper practice and mock exams, and, in Jamaica's case in 2026, direct disruption from Hurricane Melissa, which affected roughly half of the island's examination centres.