AI in the Courtroom: What the Chief Justice's New Practice Direction Means for Lawyers and Clients

Artificial intelligence has moved from novelty to necessity in legal practice almost overnight. Tools such as ChatGPT, Claude, Microsoft Copilot and Google Gemini are now used to draft correspondence, summarise documents, and even assist with legal research. The judiciary of Trinidad and Tobago has taken formal notice of this shift. On 7th July, 2026, Chief Justice Ronnie Boodoosingh signed a Practice Direction on the Use of Generative Artificial Intelligence (AI), published in the Trinidad and Tobago Gazette on 9th July, 2026 and taking effect from 8th July, 2026.

This is the first formal, binding guidance from our courts on how AI may — and may not — be used by attorneys, litigants in person, witnesses, experts and judicial officers. It applies across the board, to open-source and closed-source AI tools alike, and it comes with real consequences for those who get it wrong.

Below, we break down what the Practice Direction actually requires, the risks the Chief Justice specifically flagged, and — just as importantly — what clients themselves need to understand about how AI touches their own legal matters.

What the Practice Direction Covers

The Practice Direction is not a ban on AI. It takes a measured position: technology can assist the administration of justice, but court proceedings must remain human-controlled and human-driven. Every document that goes before a court, and every judgment that comes from the bench, must remain fundamentally the product of a person's own judgment — not an algorithm's output.

The guidance draws a careful distinction between different categories of court documents, because the risk of AI misuse is not the same for a legal submission as it is for a sworn affidavit.

The Risks the Chief Justice Identified

Before setting out the rules, the Practice Direction names the specific dangers court users need to be alert to. In summary, these include AI systems producing convincing but fabricated case citations and legislative references; results that are only as good as the data the tool was trained on, which can carry embedded bias; the possibility that information typed into a public AI chatbot could be retained and surfaced to other users; the absence of any real guarantee of confidentiality or privilege once material is submitted to an open, public AI tool; and the risk that AI training data may itself have been sourced in breach of copyright.

Anyone who has followed AI adoption in courts internationally will recognise the first risk in particular — fabricated case law, sometimes called "hallucinated" citations, has already led to sanctions against lawyers in several jurisdictions. Trinidad and Tobago's judiciary is clearly moving to head that off before it happens here.

The Rules, Document by Document

Pleadings and applications. These are legal documents that sets out a party's facts and relief they seek. AI may be used on a limited basis, in a manner comparable to using a precedent book. However, the party signing the pleading remains personally responsible for verifying that the facts stated are correct.

Affidavits and witness statements. These are a party's evidence. This is where the Practice Direction draws its firmest line. AI must not be used to generate the substantive content of an affidavit or witness statement — it may only assist with formatting, spelling and grammar. The words must remain the deponent's own, reflecting what that person actually knows and can say in their own language.

Submissions on law and facts. AI may assist, with care, in preparing chronologies, summaries, oral and written submissions, and speaking notes. But every citation, quotation, and legal authority referenced must be independently checked to confirm it exists, is accurate, and is actually relevant to the case. The responsibility for the submission's content rests entirely with the attorney or litigant who filed it — using AI does not shift or reduce that responsibility in any way, and this applies equally to litigants representing themselves.

Expert reports. The substance of an expert's opinion must remain the expert's own. AI may only be used for supporting tasks — summarising material, generating tables, or producing visual aids, indexes and chronologies.

Other permitted uses. AI may generate chronologies and indexes and assist in summarising or reviewing documents and transcripts, always subject to the user verifying accuracy.

Confidentiality: A Warning Worth Repeating

The Practice Direction is explicit that court users and experts must not input sensitive, confidential or privileged information into open-source AI tools. This is one of the most practically important points in the entire document. Many free, publicly available AI chatbots do not offer the confidentiality guarantees a law firm's internal systems provide, and material entered into them can, in some circumstances, be used to inform responses given to other users entirely unconnected to the case.

A New Disclosure Requirement

Perhaps the most significant practical change is this: where AI has been used in preparing a court document, that document must now carry a short disclosure statement at the end. It must name the specific AI tool and version used, briefly describe how it was used, and confirm that every legal authority cited has been personally verified by the person filing the document. This turns AI use from something quietly assumed into something that must be formally declared and stood behind.

Rules for Judges and the Courts Themselves

The Practice Direction does not only regulate lawyers and litigants — it applies to the judiciary too. Judicial officers may use AI to help prepare summaries of fact and evidence, chronologies, and to help identify the issues in a matter. But AI must play no role in the reasoning behind a judgment, in weighing or assessing evidence, or in editing or proofreading a draft judgment. A Court of Appeal judge who uses AI in preparing a draft judgment must disclose this to the other judges on the panel, and Judicial Research Counsel must disclose their AI use to the judicial officer they support. Judgments, the Practice Direction insists, must remain the personal work product of the judge who delivers them.

Consequences for Misuse

Judicial officers now have express authority to sanction the misuse of AI in proceedings. This can include striking out a document (or part of one), triggering disciplinary proceedings, and making adverse costs orders against the party responsible. In serious cases, the Practice Direction warns, misuse of AI could even amount to contempt of court. This is not a symbolic gesture — it puts real teeth behind the requirement that documents filed with the court be properly verified.

The Pros and Cons of AI in Legal Practice

AI is not going away, and used properly it offers genuine benefits. It can speed up the drafting of routine documents, help organise large volumes of material into chronologies and indexes, and make legal information more accessible to people who cannot afford extensive legal fees — including litigants in person navigating the court system without a lawyer.

But the drawbacks are equally real. AI tools can produce content that sounds authoritative while being factually wrong, including citing cases or legislation that simply do not exist. They can absorb and reflect biases present in their training data. They offer no guarantee of confidentiality once information is typed into a public tool. And there is a subtler risk too: over-reliance on AI can erode the independent judgment and rigorous verification that legal work has always demanded — precisely the qualities the Practice Direction is designed to protect.

The Challenge for Lawyers

For attorneys, this Practice Direction formalises what careful practitioners were already doing informally: treating AI output as a first draft that must be checked, never as a finished, reliable product. It also creates new administrative obligations — the disclosure certificate, in particular — that firms will need to build into their document preparation workflows. Firms that use AI-assisted drafting will need clear internal protocols on which tools are approved for use, what categories of information may never be entered into them, and who is responsible for verifying citations before a document is filed.

The Challenge — and the Risk — for Clients

Clients face a less obvious but equally important set of risks. A growing number of people, particularly litigants in person, are turning directly to AI chatbots to draft their own court documents, letters, or even to try to understand their legal position before ever speaking to a lawyer. The Practice Direction is a useful reminder that this carries real danger: an AI-drafted document is not automatically correct, complete, or suited to Trinidad and Tobago law, and the person who files it is fully responsible for its accuracy regardless of how it was produced.

There is also a confidentiality dimension clients should be aware of. Typing the details of a private legal dispute, a family matter, or sensitive financial information into a free, public AI chatbot means that information leaves your control the moment you hit send. Once it is in a public tool, you generally cannot verify how it is stored, retained, or used.

How Clients Should Approach AI Use in Their Matters

• Do not rely on AI-generated documents as a substitute for legal advice. An AI chatbot cannot advise you on Trinidad and Tobago law with the accuracy or context a qualified attorney can, and any document it produces must still be reviewed by a professional before it is relied upon.

• Be cautious about what you share with public AI tools. Avoid entering sensitive personal, financial, or case-specific details into free or open-source AI chatbots.

• Ask your attorney about their AI practices. You are entitled to understand how a firm handling your matter uses AI tools and what safeguards are in place to protect your confidential information.

• Understand that responsibility does not disappear. Whether a document was drafted by a lawyer, a litigant in person, or with AI assistance, the person who signs and files it remains responsible for its accuracy before the court.

• Treat AI-generated legal information as a starting point, not an answer. If you have used AI to research your situation, bring those questions to a consultation rather than acting on the output directly.

  
  

Looking Ahead

The Practice Direction is explicit that it will be reviewed periodically as AI technology continues to evolve, and it calls on the Law Association of Trinidad and Tobago and the Hugh Wooding Law School to train the next generation of practitioners in its responsible use. For now, it sets a clear marker: AI is a tool that may assist the practice of law in Trinidad and Tobago, but it cannot replace the judgment, verification, and accountability that the administration of justice depends on.

Frequently Asked Questions

Does this Practice Direction ban lawyers from using AI? No. It permits AI use in defined circumstances — such as chronologies, summaries and limited drafting assistance — while prohibiting its use to generate the substantive content of affidavits, witness statements, and expert opinions.

Can I use ChatGPT or a similar tool to draft my own court documents if I don't have a lawyer? You are not prohibited from doing so, but you remain fully responsible for verifying that everything in the document — every fact, citation, and legal reference — is accurate. The Practice Direction makes clear that litigants in person are held to the same standard as attorneys in this respect.

What happens if AI-generated errors end up in a court filing? Judicial officers can strike out the document or part of it, refer the matter for disciplinary action, order costs against the responsible party, and in serious cases, treat the misuse as contempt of court.

Is it safe to discuss my case with a free AI chatbot? Caution is strongly advised. Information entered into open, public AI tools may not be protected by any duty of confidentiality, and the Practice Direction specifically warns against inputting sensitive or privileged information into such tools.

Brent Kevin Ali is a civil litigation attorney with more than 20 years experience. If you have questions about how this Practice Direction affects a matter you are currently involved in, or if you would like advice before filing a document you have prepared with AI assistance, contact Brent Kevin Ali at 868-235-6016 or through www.brentkali.com to schedule a consultation.