On 11 May 2026 the Caribbean Examinations Council reaffirmed that students may use AI tools in school-based assessments across CSEC, CAPE and CCSLC. It is one of the most consequential education decisions the region has made in years, and it arrives with risks the region is poorly prepared to manage.
The timing matters. ECLAC forecasts Caribbean growth outside Guyana at 1.7 percent in 2026, held back by softer US tourism demand, high energy costs and climate shocks. Hurricane Melissa's landfall in Jamaica in October 2025 caused damage and losses assessed at US$12.2 billion, equal to 56.7 percent of the country's GDP. About 500,000 people work in Caribbean tourism, a sector that has lost its post-pandemic momentum. The students now allowed to use AI in their coursework will enter that economy within a few years.
What CXC's AI rules require from students and teachers
The framework rests on fairness, human judgement and academic integrity. In practice:
- a student who uses AI in an SBA runs an originality check and includes the report with the submission;
- the teacher completes a CXC AI Disclosure Form;
- AI detection software is not used as the only basis for judging a student's work.
I think this is the right decision. Today's CXC candidates will work in jobs where AI is a routine tool, and coursework that pretends otherwise prepares them for a workplace that no longer exists. Moving the question from whether students may use AI to how they should use it also puts the weight on documentation and checking AI output, which every career will need.
The rules permit AI. Teaching students to use it well falls to curricula, teacher training and devices, and none of those comes with the policy.
Why the regional economy needs students who can use AI
The Caribbean Development Bank's 2026 outlook describes modest growth amid global uncertainty. The pressures underneath are specific: tourism exposed to the US economy, electricity that in many islands costs at least twice what US households pay, recurring climate damage and a steady loss of trained people to North America and the UK.
Guyana's growth comes from oil. The rest of the region needs higher productivity in services, farming and health without matching rises in labour cost, and people who can build and run AI systems are part of that. AI alone will not do it. A region where AI skills stay shallow will keep buying expertise from abroad.
AI tutoring for students without private lessons
A student in rural St Elizabeth, on Nevis or in rural Belize does not have the teachers, extra lessons or materials available to a student in Port of Spain or Kingston. An AI tutor set up for the CSEC syllabus can offer worked examples, adapt to the student's pace and spot gaps at any hour, with no travel or lesson fees. Such tools exist now. What the region lacks is devices and connectivity in the right homes, teachers trained to use the tools, and a curriculum that says how they fit.
Two other uses are close behind. Students miss regional scholarships and tertiary places because the information is scattered and the applications are complex, and an AI adviser loaded with that information could help. In technical and vocational training (IT, health support, construction, agri-processing), AI tools can shorten the time it takes to reach a working standard.
Faster returns: tourism, food, health and credit
Education pays back over years. These sectors can see results sooner:
- Tourism: small hotels and guesthouses can use AI for pricing, guest personalisation, maintenance prediction and cheaper digital marketing, so they compete less on sun and sea alone.
- Agriculture: the region imports much of its food. Rainfall, pest and soil models now work at small-farm scale. The Caribbean Agricultural Research and Development Institute (CARDI) has the mandate; rural connectivity is the gap.
- Health: AI tools can extend scarce specialists by flagging high-risk patients earlier. Barbados and Jamaica have enough health data infrastructure to start; procurement and political will are the barriers.
- Credit: alternative credit scoring from payment and transaction records can bring informal workers into formal finance, as it has in Sub-Saharan Africa and Southeast Asia. Caribbean versions need training on Caribbean data.
Risks: the device gap, integrity, data and bias
The device gap. Without devices, connectivity and teacher training in under-resourced schools, the policy gives the biggest advantage to students who already have the most. That would reverse its purpose.
Integrity. Disclosure and originality reports are a reasonable start. Teachers will need regular training, every year, on what AI-assisted work looks like and how to find the student's own contribution in it.
Data. Students and teachers using global AI platforms send their data to companies outside the region. Without regional rules, the Caribbean supplies data without much say in how it is used.
Bias. Tools trained mostly on North American and European material give examples built around US incomes and job markets. A tutor can be technically right and still wrong for a Caribbean student. Ministries should test tools on local syllabus content before recommending them.
I do not yet know how CXC will moderate SBAs where AI use was disclosed but heavy. That is the part of the policy most likely to change after the first marking cycle.
What to do next
- Ministries of education: count, school by school, the CXC candidates without a working device and home internet, and publish a funded plan to close that gap by the 2027 SBA deadlines.
- Principals and heads of department: run a half-day session before the next SBA cycle on the CXC AI Disclosure Form, originality checks and how to question a student about their draft.
- Teachers: require students to keep dated drafts or version history for every SBA, so AI-assisted work can be traced back to the student's own thinking.
- Students: keep a short log of each AI prompt and what you changed in the output. It makes disclosure easy and shows your own contribution if questioned.
- CARICOM and national data protection offices: agree a common position on student data in AI tools used by schools, starting from the Data Protection Acts already passed in Jamaica and Barbados.
- CDB and national governments: put a line for AI curriculum and teacher training into the next education loan or grant, costed per school.
Frequently Asked Questions
Does CXC allow AI in the written exam papers too?
No. The AI provisions cover school-based assessments, the coursework completed during the year. Timed written papers are still sat under supervised examination conditions, where phones and other devices are not allowed. Students need to be able to do the work unaided on exam day.
Are students old enough to use the main AI tools?
Check each tool's terms before a class uses it. OpenAI requires ChatGPT users to be at least 13, with parental permission under 18, and Anthropic's consumer Claude app is for users aged 18 and over. Schools should list which tools are approved for which age groups and get parental consent where the terms require it.
How can a teacher check a student's AI-assisted SBA without detection software?
Ask for evidence of how the work was made. Dated drafts, version history in Google Docs or Word, the student's prompt log and a short conversation in which the student explains choices in their own words show far more than a detector score. A student who cannot explain a paragraph they submitted needs a follow-up conversation.
What happens if a student uses AI and does not disclose it?
Undisclosed use falls under CXC's academic integrity rules, and the teacher cannot honestly sign the disclosure form for that work. Read the current CXC AI policy and SBA guidelines on the CXC website for the exact consequences, which the school's CXC coordinator can also explain. Disclosing is simple, so there is no advantage in hiding it.
Which Caribbean governments have set up AI-specific ministries or laws?
Trinidad and Tobago's government created a Ministry of Public Administration and Artificial Intelligence in 2025. Jamaica has a National AI Task Force and the Data Protection Act 2020, in force since December 2023, and Barbados has its Data Protection Act 2019. No CARICOM-wide AI or data framework exists yet, so rules on student data differ by country.