There is a common assumption that the difference between a legal-AI tool built for the Western market and one that serves the Arab region is a matter of language and nothing more: add Arabic support, the thinking goes, and the problem is solved.
Anyone who practises law in the Gulf knows the issue runs far deeper. Language is only the visible layer of a structural difference that reaches into the core of how law is written, interpreted, and applied. When tools built on English common law are applied to Arabic statutory text, three recurring errors appear, each deeper than the last.
The first error: reading for precedent where the text governs
Common-law systems reason from accumulated case law. Each judgment builds on those before it, and the legal rule takes shape over time through the rulings of the courts. The major legal-AI tools were trained on exactly this logic: find the precedent, weigh one ruling against another, derive the rule from the accumulation of judgments.
Civil-law systems, which govern most Arab states, work in a fundamentally different way. The primary source is codified statute. The judge applies the text rather than making it, and the role of judicial interpretation remains bounded and constrained by the legislative framework. A tool that instinctively searches for “the precedent” in an environment governed by the text is not making a passing mistake; it misidentifies the source of legal authority itself. It asks the wrong question before it begins searching for an answer.
The second error: treating the Arabic text as a translation
In many states across the region, the law requires Arabic for official transactions and filings. This means the Arabic text is not a translated copy of a document whose original is in English. It is the operative document itself, the text relied upon before the courts.
Yet international platforms are built on an English-first core, with Arabic added as a later layer. That ordering inverts the actual hierarchy of legal work: what is a binding original is treated as a secondary translation, and what is secondary is treated as the original. This is the gap a regionally built platform is designed to close. Companies like Oqood AI, whose systems are built for bilingual, civil-code practice from the ground up rather than retrofitted from an English-first core, treat the Arabic text as what it legally is, the operative document, not a translation layered over it. The distinction is not cosmetic. It is the difference between correctly identifying which text is the authority and which is merely persuasive, a distinction that can prove costly when a dispute reaches court.
The third error: ignoring that legal terms are not neutral
A legal term is not a word in a dictionary. It is a concept loaded with its legislative context and judicial history. Many terms in Arab civil law have no precise equivalent in English common law, because they belong to a fundamentally different legal structure.
When a tool translates an Arabic term into the nearest English word, or the reverse, it carries the word across and drops the meaning. The output may look correct to a non-specialist, but before a lawyer or a judge it reveals a gap in understanding: a concept moved from one system to another without its framework. In legal work, that gap is the difference between a tool that can be relied upon and one that becomes a source of risk.
Why this matters now
These three errors are not raised as theoretical criticism. They bear on a practical decision that law firms in the region are making today. As AI adoption accelerates across the legal profession globally, firms in the Gulf face a choice between powerful tools designed for another legal system and another language, and tools built from the ground up for the environment in which they actually work.
The question, at its core, is not which tool is more sophisticated, but which one understands the law as it is practised here. A tool that reads for precedent where the text governs, treats the Arabic original as a translation, and moves terms without their framework does not merely provide an incomplete service. It offers a mistaken understanding of a law written in a logic it was never designed to read. And for a profession built on precision, that is a difference that matters.
