The rapid proliferation of artificial intelligence tools in academic spaces has pushed the Caribbean Examinations Council (CXC) to roll out updated, stricter protocols for School-Based Assessments (SBAs) ahead of the June 2026 examination cycle. The new policy was announced by Magalie Celestine, CXC’s local registrar for Dominica, during a public presentation of 2025 examination results held Monday at the Dominica Public Service Union Building, where she addressed attending educators, parents, students, and members of the regional press.
Under the revised framework, all students completing SBAs must develop their work in full alignment with CXC’s newly updated official Standards and Guidelines for artificial intelligence integration in academic assessments. A key new requirement mandates that every SBA submission undergo automated and manual originality screening to detect uncredited AI-generated content. Additionally, both participating students and their supervising teachers are required to sign formal honesty declarations confirming the work meets academic integrity standards, which are submitted alongside the completed SBA.
Celestine openly acknowledged that these additional anti-plagiarism and AI oversight measures have created a heavier administrative and workload burden for both examination candidates and teaching staff. Compared to previous examination cycles, the revised SBA submission process is far more rigorous and time-intensive for all parties involved.
Despite these growing growing pains, Celestine noted that the current SBA model for most core subjects is nearing its end. Starting from the June 2027 examination cycle, SBA submissions will no longer be required for 13 Caribbean Secondary Education Certificate (CSEC) subjects and 17 Caribbean Advanced Proficiency Examination (CAPE) subjects. This phase-out is a core component of CXC’s long-term strategy to overhaul its assessment practices to address the shifting landscape of academic work, driven largely by the expanding influence of generative artificial intelligence.