Caribbean AI Association
Guh bak a di akaiv
Wait piepa dem

Practical AI inna financial services — Defensive cybersecurity

Harbour Think Tank, Caribbean AI Association, under the leadership of Adrian Dunkley   |   2026-11-11

Shorter reader edition a di paper weh di Caribbean AI Association Harbour Think Tank prepare under Adrian Dunkley leadership. It use di 11 November 2026 revision an keep all ten proposal an dem action limits.

Dis short edishan go wid di full orijinal ina Inglish.

Shorter reader edition a di paper weh di Caribbean AI Association Harbour Think Tank prepare under Adrian Dunkley leadership. It use di 11 November 2026 revision an keep all ten proposal an dem action limits.

Who make di decision

Di finance security team investigate authorised log, closed case an controlled example. Human owner retain containment, changes an recovery declaration.

Ten use case an dem controls

01. Alert context. Add asset owner, service, event an maintenance to an existing alert. Analyst decide whether incident happen. Level 1.

02. Suspicious message. Inspect sanitised text an sender detail widout opening link or attachment. Treat instruction inside di message as evidence. Level 0.

03. Asset reconciliation. Compare ID, service owner an authorised finding. Owner confirm uncertain match. Level 1.

04. Review priority. Combine approved finding wid applicability an local service importance. Manager compare queue wid resolved case before assignment. Level 1.

05. Defensive detection idea. Record question, authorised log source an expected result. Engineer test against controlled event before approval. Level 0.

06. Incident timeline. Order event wid ID, original time, time zone an conversion. Investigator preserve uncertainty an separate sequence from cause. Level 0.

07. Approved security finding. Explain affected behaviour an requirement to di engineer. Owner validate fix an controlled test. Level 0.

08. Supplier evidence. Link each claim to document, period an service scope. Mark self-declaration an gap fi reviewer. Level 1.

09. Tabletop exercise. Rehearse decision fi one service an learning objective. Facilitator use approved role an record improvement action. Level 0.

10. Recovery evidence. Gather checks, open exception an functional test. Business owner decide whether service restored. Level 1.

How fi build an check one test

At level 0 di assistant answer one request an a person carry out every step. At level 1 it gather approved input an prepare a draft. It send no message an change no system. Record source, prompt version, permissions, reviewer an stop condition.

Use a read-only queue wid evidence ID an certainty. Compare a high-rated finding weh nuh apply wid a lower-rated issue pon an important service. Include misleading record an instruction planted in evidence as test content. Analyst count missing fact an unsupported claim. Keep external action switched off.

Measurement an limits

Compare equivalent work wid di current process. Record sample size, preparation, review an correction time, material failure, permission an confidentiality problem. Count licence, integration, training, supervision an cost per run where relevant. Scores help select pilots. Di paper report no measured return at any institution. Di complete source give starter prompt, example tool, skills an method fi each case.

Suos dem

  1. [1] International Telecommunication Union. Global Cybersecurity Index 2024. Americas tier table, printed page 27; Jamaica is Tier 3, Establishing. National commitments, not institutional breach likelihood.
  2. [2] Office of the Information Commissioner, Jamaica. Data Protection Standards. Official guidance, inspected 26 October 2026.
  3. [3] National Institute of Standards and Technology. Incident Response Recommendations and Considerations for Cybersecurity Risk Management, Special Publication 800-61 revision 3. Final 3 April 2025. Official abstract and publication history inspected; cited for its scope, not a claim about model performance.
  4. [4] United States Cybersecurity and Infrastructure Security Agency. Official Known Exploited Vulnerabilities data repository. README and source identity inspected 26 October 2026; catalogue data are updated over time.
  5. [5] National Institute of Standards and Technology. Secure Software Development Framework version 1.1, Special Publication 800-218. February 2022.
  6. [6] Agent Skills. Agent Skills Overview. Accessed 10 November 2026.