Jason Nathu considers the impact and importance of the Judiciary’s Practice Direction on the use of AI in the legal system, brought into force in July 2026.
Artificial intelligence is changing the way professionals work. Doctors, accountants, architects, engineers; none have been spared, and lawyers are no exception.
Across the Caribbean and around the world, attorneys-at-law are using AI tools to conduct legal research, draft documents, prepare submissions, and manage cases at a pace that was unimaginable even five years ago.
But with that speed comes risk. And in Trinidad and Tobago, the courts have now stepped in with clear rules about how AI may and may not be used in legal proceedings.
On 9th July 2026, the Practice Direction on the Use of Generative Artificial Intelligence was gazetted, coming into force on 8th July 2026. It matters not just for lawyers and judges, but for anyone who might one day find themselves in court, whether as a party, a witness, or a litigant in person navigating the system without legal representation.
What is Generative AI, and why does it matter in a courtroom?
Generative AI (or Gen AI) refers to tools that can create new content: text, images, summaries. These tools work by learning patterns from vast amounts of existing data.
Tools like ChatGPT, Google Gemini, Microsoft Copilot, and Harvey are now commonly used in legal practice. Some are general-purpose platforms; others are specifically designed for lawyers.
The appeal is obvious. These tools can produce a polished first draft of a legal submission in seconds. They can summarise a lengthy judgment in plain language. They can generate a chronology of events from a stack of documents that would take a paralegal hours to work through.
What they cannot do is guarantee accuracy. This is critical.
The hallucination problem
AI tools have a well-documented tendency to fabricate. In the AI world, this is called “hallucination.” The system produces content that looks entirely credible. It includes case names, citations, and page references. Yet the content simply does not exist.
This is not a theoretical concern. It has happened here in Trinidad and Tobago.
In Nexgen Pathology Services Ltd v Darceuil Duncan CV 2023-04039, the High Court of Trinidad and Tobago found that counsel had relied on multiple non-existent cases. None of these cases could be found in any recognised legal database.
The Honourable Mr. Justice James referred the matter to the Disciplinary Committee of the Law Association of Trinidad and Tobago, describing it as a serious breach of professional ethics. His message to those tempted to let AI do their thinking for them was pointed: “Rest assured, the intelligence of this Court is not artificial.”
The same judgment acknowledged that AI tools are not going away. As the Court put it, “digital tools including AI and internet-based platforms are increasingly common and valuable in legal research; indeed, this Court itself makes use of such tools where appropriate.” The problem is not the technology. The problem is uncritical reliance on it.
The courts have been equally clear about what responsible use looks like. In the Uk case of Ayinde v London Borough of Haringey [2025] EWHC 1040, the Court stated that lawyers “who use artificial intelligence to conduct legal research notwithstanding these risks have a professional duty therefore to check the accuracy of such research by reference to authoritative sources, before using it in the course of their professional work, to advise clients or before a court.”
That duty of verification sits at the heart of the new Trinidad and Tobago Practice Direction.
What does the Practice Direction actually say?
The Practice Direction draws clear lines between what is permitted and what is not.
For pleadings and applications, AI may be used in a limited way. This is similar to how a lawyer might draw on a precedent book to structure a document. But the lawyer signing the document remains personally responsible for verifying that every fact is accurate.
For affidavits and witness statements, AI is prohibited from generating the content. These documents exist for a specific purpose. They are the witness’s own account of facts, beliefs, and opinions, given under oath. Allowing a machine to generate those words, even if a human then approves them, fundamentally undermines the integrity of sworn evidence.
The Practice Direction is clear that affidavits and witness statements must remain the product of the deponent or witness, as appropriately guided by their legal adviser.
For submissions on law and facts, AI may be used with care. But every case cited, every quotation from statute or authority, and every legal principle invoked must be independently verified by the lawyer. The tool can assist; the judgment must remain human.
For expert reports, AI may not be used to draft the substance of the report or the expert’s opinion. The expert’s view must be their own.
For judicial officers, AI may assist with chronologies, summaries, and identifying issues. But it must play no role in formulating the reasons for a judgment, in the assessment of evidence, or in editing a draft judgment.
And for all court documents where AI has been used, a certificate of disclosure is required. This certificate must briefly describe how the tool was used and confirm that all legal authorities have been verified.
The ethics dimension
The Practice Direction connects directly to a lawyer’s core professional obligations.
The duty of competence requires lawyers to understand the tools they use. Using Gen AI without understanding how it works, where it fails, and how to verify its output is not a shortcut. It is a risk to the client and to the lawyer’s professional standing.
As Mata v Avianca Inc., one of the most widely cited AI hallucination cases globally, made plain: “AI, while a promising tool, is not a replacement for human responsibility and oversight. A lawyer is answerable for their research, arguments, and representations under their core duties to the court and to their client.”
The duty of confidentiality is also engaged. Lawyers who input sensitive client information into a public AI platform may be unknowingly exposing privileged material. The Practice Direction specifically warns court users and experts to refrain from inputting sensitive, confidential, or privileged information into open-source Gen AI tools.
The duty to the court, which sits above even the duty to the client, means that submitting AI-generated content that has not been properly verified is not just sloppy practice. In serious cases, it could amount to contempt of court.
A word for self-represented litigants
If you are representing yourself in court without a lawyer, you might be tempted to use AI tools like ChatGPT, Gemini, or Claude to help draft your court documents. The tools are free, they are easy to use, and they produce text that looks professional. The temptation is understandable.
But the Practice Direction applies to you just as it applies to lawyers. The same rules about accuracy, verification, and disclosure apply. And the same risks apply as well.
If you use AI to generate a legal submission or pleading, you must verify every case cited, every quotation from statute or law, and every factual claim.
You cannot simply copy what the AI produces and file it with the court. If the AI has hallucinated a case, or misquoted the law, the court will hold you responsible. The fact that you are representing yourself does not excuse you from the obligation to get it right.
The Practice Direction also requires that if you have used AI to prepare court documents, you must disclose that fact to the court in writing. This is not optional. The court needs to know.
More fundamentally, AI cannot replace judgment. It cannot understand the unique circumstances of your case. It cannot weigh the facts against the law in a way that takes account of your particular situation.
What AI can do is help you organise your thoughts, draft a first version of a document, or check your grammar. But the final work product must be yours, and you must understand it, believe in it, and be prepared to stand behind every word.
If you do not have the resources to hire a lawyer, there are other options. The Legal Aid Clinic at Hugh Wooding Law School, community legal centres, and bar association referral services exist precisely to help people who cannot afford representation. These resources may take time to access, but they are far safer than relying on an AI tool that does not understand your case.
This is not anti-technology
It is worth being direct about what this Practice Direction is not. It is not a ban on AI. It is not a statement that technology has no place in the law. The rules explicitly permit AI assistance in a range of tasks: drafting, summarising, generating chronologies, and supporting research.
The Judiciary recognises that these tools are here to stay, and that used responsibly, they can genuinely improve access to justice and the efficiency of legal proceedings.
What the Practice Direction insists on is that AI remains a tool rather than a replacement for human judgment. The lawyer, the expert, the judicial officer. Each remains personally responsible for every word submitted to the court in their name.
Trinidad and Tobago is, in fact, one of several Caribbean jurisdictions to issue guidance of this kind, and it is worth understanding where it fits in that sequence. The Caribbean Court of Justice was the first in the region to act, issuing its Practice Direction on the use of generative AI tools on 14th February 2025.
The Turks and Caicos Islands followed on 4th August 2025 with its own Practice Direction, and Jamaica issued its version on 17th September 2025 under Chief Justice Bryan Sykes.
The Bahamas issued its Practice Direction on 5th November 2025, prompted in part by a probate matter in which fabricated case law had been placed before the court. Trinidad and Tobago’s Practice Direction, gazetted on 9th July 2026, is the most recent addition to this growing body of regional judicial policy.
No other Caribbean jurisdiction appears to have issued equivalent guidance at this time, though that is likely to change as the technology becomes further embedded in legal practice across the region.
What this means for you
If you have a legal matter before the courts, whether you are represented by a lawyer or navigating the system yourself, you have a right to know that the documents filed on your behalf are accurate, verified, and genuinely the work of a human being who has applied professional judgment to your case.
The Practice Direction gives practical effect to that expectation. It places the responsibility squarely where it belongs: with the court user, not the machine.
AI can help a lawyer work smarter. It cannot replace the obligation to work carefully.
About the author
Jason Nathu is an attorney-at-law admitted to practice in Trinidad and Tobago (2006) and Guyana (2023), with nearly two decades of experience in civil litigation, intellectual property, family law, conveyancing, and estate matters.
He currently serves as Senior Tutor II and full-time faculty member at the Hugh Wooding Law School (HWLS), in the Legal Aid Clinic which provides legal representation to persons of limited means, while simultaneously providing students with rigorous, immersive experiential training and learning.
At HWLS, Jason also supervises the Human Rights Law Clinic and the Innovation, Technology & Entrepreneurship Clinic, and lectures in the Law Office, Management and Technology course. He frequently appears before the Court of Appeal, the High Court, and the Equal Opportunity Tribunal and has acted in several significant matters involving copyright and discrimination issues.
Jason is deeply involved in legal governance and professional development within the region. He is a former member of the Council of the Law Association of Trinidad and Tobago and continues to serve on its Information Technology, Human Rights, and Publications Committees. He also serves as Vice-President of the American & Caribbean Law Initiative (ACLI) and is an academic member of the International Trademark Association (INTA).
Academically, Jason holds a BA in History and an LLB from the University of the West Indies, a Legal Education Certificate from the Council of Legal Education, and a Master of Laws (LLM) in Intellectual Property Law from the University of London.
Jason is an active researcher, presenter, and commentator on the intersection of law, technology, and society, with particular emphasis on artificial intelligence, intellectual property, and access to justice in small jurisdictions.
He remains committed to strengthening experiential learning, public legal education, and the development of innovative pathways for future Caribbean lawyers.
