JAMAICA Trends and Developments Contributed by: Marc Ramsay, Samantha Moore and Chantelle Biersay, Ramsay & Partners
and correspondence, but it sets strict conditions on how these tools may be used. Attorneys remain fully accountable for the contents of all documents filed. Any submission prepared with GenAI, whether in part or in whole, must be independently verified for factual accuracy and legal soundness. The Court makes it clear that responsibility cannot be shifted to technol - ogy. The duty of candour and diligence continues to rest with counsel, regardless of the sophistication of the tool employed. Prohibited uses and protection of evidence integrity Certain uses of GenAI are expressly prohibited. Attor - neys may not rely on it to draft affidavits, witness statements or expert reports. Nor may it be used to generate or alter evidence, or to produce fabricated case law, statutory provisions or legal principles. The creation of avatars, images or likenesses for participa - tion in virtual hearings is also forbidden. The scope of these restrictions makes plain the Court’s intention to preserve the authenticity of evidence and the integrity of testimony. Judicial scrutiny and evidentiary standards The Practice Direction also introduces a formal disclo - sure obligation. Where any portion of a court docu - ment is prepared with the assistance of GenAI, the filing party must submit a declaration in the prescribed form. This declaration identifies the tool used and con - firms that all citations and representations have been independently verified. Attorneys should treat this requirement with the same seriousness as the signing of pleadings or affidavits, since failure to comply may lead to documents being struck out, costs sanctions, contempt proceedings, or referral to the Disciplinary Committee of the General Legal Council. In less than obvious cases, determining whether an attorney has failed to disclose the use of GenAI will be difficult, as current AI detection tools such as GPTZero, Copyl - eaks and Turnitin may flag generative text, but their accuracy rarely exceeds 80% in practice, with para - phrasing or minor edits often defeating detection and false positives posing real risks in legal writing. Judicial officers will assess AI-assisted submissions strictly against the same evidentiary and procedural rules that apply to all filings. The Court has made clear
it will not adjust or relax these standards to accommo - date the use of new technology. This position signals caution: while the judiciary permits the use of GenAI, it will not extend any indulgence if the material contains errors, fabricated authorities or other defects. Attor - neys must therefore treat every AI-assisted document as though it will be scrutinised with no margin for error, exactly as if it had been drafted entirely without tech - nological support. Practical uncertainties and emerging distinctions between tools The framework is both enabling and cautionary. On the one hand, it confirms that the use of GenAI is per - missible in Jamaican courts. On the other, it highlights risks that demand further thought from the profession. The prohibition on uploading confidential or privileged information to unsecured platforms is straightforward, yet the status of enterprise-grade or closed-system AI solutions remains uncertain. While the requirement to declare the use of GenAI is unambiguous, the extent of disclosure when tools are used merely for stylis - tic refinement, as opposed to substantive drafting, is less clear. For example, the use of unpaid general- purpose foundation models (LLMs) such as ChatGPT and Gemini is vastly different from paid models. Fur - thermore, the use of LLMs differs significantly from that of domain-specific applications, such as Vincent by vLex, Lexis+ AI, Harvey or the more familiar tool Grammarly. It is likely that attorneys who can disclose the use of credible, source-grounded, specialised legal GenAI tools will have an advantage, practically and perceived. Recommended safeguards for practitioners For practitioners, the prudent course is to imple - ment internal safeguards. Every output from GenAI should be reviewed against primary sources and veri - fied for accuracy. Confidential data should never be processed through unsecured systems. A clear jus - tification record should be maintained, documenting why GenAI was used, how accuracy was confirmed and what protective measures were applied. Above all, attorneys must remember that the Court expects strict adherence to the Practice Direction rather than a creative interpretation.
235 CHAMBERS.COM
Powered by FlippingBook