ADGM Courts Registrar and CEO Linda Fitz-Alan told the ADGM Global Media Forum, in remarks reported on 21 October 2026, that the courts are close to launching AI-powered services. Khaleej Times reported that the tools are meant to help users navigate court procedure and complete documents. They are also expected to support faster claim reviews and the preparation of court orders and judgments. 'We are very close to launching this, expected to be out by the end of the year,' she said. The report describes the approach as cautious and controlled because of the responsibilities of judicial decision-making.
No new law sits behind the announcement. The UAE still has no standalone AI statute, and a September 2025 Clyde & Co analysis says the same about AI in dispute resolution. ADGM, a common-law free zone in Abu Dhabi, is deciding by institutional practice how AI enters adjudication. That is a reasonable way to start, provided the courts publish enough about how they are doing it.
The strongest case for waiting for a statute
Courts are different from other AI deployments. A tool that drafts a hearing order or triages a claim can shape outcomes even when a judge signs the final text. Automation bias, hallucinated authorities and opaque training data are real risks. A statute would give litigants enforceable rights: notice that AI was used, a route to challenge it, and a clear line that decisions stay with a human judge. Without that, safeguards depend on the goodwill of the institution that benefits from the efficiency. That is a serious argument, and the courts' own caution implicitly accepts part of it.
Why institutional practice is the better starting point
The argument has limits. A general AI statute would be a blunt instrument for a narrow problem. The UAE's national AI strategy page lists, as one of five themes, a commitment to 'issue a government law on the safe use of AI'. The page does not mention courts or justice at all. A broad safe-use law drafted for the whole economy would have little to say about whether an AI tool may help prepare a court order. A court is also the party best placed to learn from its own deployment. ADGM can pilot, measure and adjust faster than any legislature, and it can do so in a jurisdiction where litigants choose the forum.
The starting scope is also modest, and it matches where the evidence for AI benefit is strongest. Form completion, procedural navigation and claim triage are administrative tasks. ADGM's electronic platform dates to 2018 and already lets users file claims, pay fees, attend hearings and receive judgments online, and parties outside the UAE can use it remotely. Adding AI assistance to a mature digital process is a smaller step than introducing AI into a paper-based system.
The workload pressure is real. The courts had recorded more than 900 cases this year by the time of the forum, and they expect about 1,200 by year-end. When caseload grows faster than judicial headcount, access to justice suffers through delay. Well-governed automation of routine steps can protect time for the contested questions that need a judge.
What the courts should publish
Institutional practice is only as credible as its transparency. Three things would make ADGM's approach durable.
- A written scope statement. The reported uses span user guidance, document completion, claim review, and preparation of orders and judgments. Those carry very different risk. Users should be told which tools do which, and that a human judge makes and signs every decision.
- Disclosure and challenge rights. Parties should know when AI assisted in preparing a court document and be able to raise concerns. Neighbouring DIFC Courts already require parties to 'declare at the earliest possible opportunity' any intention to use AI in proceedings, under Practical Guidance Note No. 2 of 2023. It also tells parties not to rely on AI-generated content without verifying it. The note governs litigants rather than the court's own tools, but it shows the region's courts can publish workable rules without waiting for legislation.
- Outcome monitoring. The courts should track error rates, reversal rates and user complaints, and say how they will pause a tool that underperforms.
I could not confirm whether ADGM Courts has published any AI practice direction. The court's website was not accessible to me, and the Khaleej Times report mentions none. If none exists, publishing one before launch is the cheapest credibility the courts can buy.
The wider UAE context
The UAE approaches AI through strategy, councils and sector programmes rather than one omnibus law. Its government AI policies page describes a UAE Council for Artificial Intelligence set up to make the country a world leader in the sector by 2031, and says the UAE is working on policies and regulations that align with future approaches to generative AI. That page also does not address judicial use. Free zones like ADGM and DIFC therefore act as regulatory laboratories, each moving at its own pace.
That model carries a fragmentation cost. Lawyers working across onshore courts, DIFC and ADGM may meet different expectations on AI disclosure and verification. But fragmentation is a manageable price for letting institutions test approaches before they are hardened into federal law. If ADGM's deployment goes well, its published practice can inform any later legislation. If it goes badly, the damage is contained and correctable, which is much harder to say of a statute.
Bottom line
Waiting for a comprehensive AI law before using AI in courts would delay modest, useful tools without making them safer. The responsible course is the one ADGM describes: a staged launch on an existing digital platform, with human judges keeping the decisions. What remains is to put the safeguards in writing. A court that publishes its rules, measures its results and keeps a human accountable for every judgment will earn more trust than any statute passed in advance of the evidence.