Civil Law and UAE Machine-Readable Law and Semantic Legal Systems

1. Introduction

Machine-readable law means legislation, regulations, judgments, contracts, and legal metadata structured so that a computer can identify, process, search, connect, and apply legal information.

A semantic legal system goes one step further. It attempts to make the meaning and relationships within legal rules computationally understandable—for example:

• identifying which rule applies to a particular fact;
• connecting a statutory provision with exceptions;
• identifying definitions;
• linking a provision to amendments;
• determining temporal validity;
• connecting a rule to its jurisdiction;
• mapping legal concepts such as good faith, public order, causation, or damages.

The UAE is particularly significant in this field because its legal infrastructure increasingly combines digital legislation, electronic transactions, digital courts, AI-assisted litigation, digital evidence and specialised technology adjudication.

However, an important distinction must be maintained:

Making law machine-readable does not make legal interpretation fully machine-determinable.

The current UAE framework supports digital processing of law, but it does not transform legal interpretation into a purely computational exercise.

2. Meaning of Machine-Readable Law

Traditional legislation is primarily written for human readers.

For example:

“A person who causes damage to another shall be liable to compensate for the damage.”

A machine-readable version could additionally encode:

• legal rule;
• actor;
• prohibited/regulated conduct;
• causation;
• damage;
• liability;
• remedy;
• exceptions;
• effective date;
• jurisdiction.

A computer could therefore represent the relationship approximately as:

Conduct → Causation → Damage → Liability → Remedy

This does not mean that the computer has understood the entire legal meaning.

It means that the structure of the rule has been represented in a format that software can process.

3. Machine-Readable Law vs Semantic Law

Example

Machine-readable system:

Article 100 → contract → effective date → 1 June 2026.

Semantic system:

Article 100 applies to this contractual relationship, subject to Article 105, but Article 105 does not apply because the transaction falls within an exception.

The second operation is much more difficult.

4. UAE Civil Transactions Law and Machine-Readable Law

The new UAE Civil Transactions Law, Federal Decree by Law No. 25 of 2025, is particularly important.

Article 1 establishes a hierarchy of legal sources:

1. legislative provisions;
2. Islamic Sharia where legislation does not provide the answer;
3. custom, provided it does not conflict with public order or public morals;
4. principles of natural law and rules of justice where no applicable rule exists.

Article 1 also states that there is no room for ijtihad where the legislative text is definitive. Article 2 refers to the principles of Islamic jurisprudence for understanding, interpreting and constructing legislative texts.

This creates a major limitation for semantic automation.

A computer can encode:

Rule A → Rule B → Exception C

but the legal system may require a deeper process of:

text → interpretation → jurisprudential principle → custom → justice

Therefore, not every legal concept can be reduced to a fixed computational rule.

5. The Core Principle

The central principle is:

Machine-readable law can represent legal rules; it cannot automatically exhaust the meaning of law.

This distinction is crucial.

A statute may contain:

• precise numerical requirements;
• deadlines;
• definitions;
• procedural conditions.

These are relatively easy to encode.

But concepts such as:

• good faith;
• abuse of rights;
• reasonableness;
• public order;
• justice;
• causation;
• foreseeability;
• proportionality;

are much harder to convert into deterministic algorithms.

6. Legal Ontologies

A legal ontology is a structured representation of legal concepts and their relationships.

For UAE civil law, an ontology could contain:

Persons

• natural person;
• legal person;
• company;
• public authority;
• association.

Obligations

• contractual obligation;
• tortious obligation;
• restitution;
• indemnity.

Events

• contract formation;
• breach;
• damage;
• termination;
• payment.

Remedies

• damages;
• specific performance;
• rescission;
• restitution;
• injunction.

Relationships

For example:

Contract

→ creates → Obligation

→ may be breached by → Conduct

→ breach may cause → Damage

→ damage may produce → Compensation

Such structures make legal databases significantly more powerful.

7. Semantic Legal Reasoning in the UAE

A semantic legal engine could theoretically receive:

“A supplier failed to deliver goods on time.”

It could identify:

Supplier

→ contractual party

Failure to deliver

→ possible breach

Delay

→ temporal event

Contract

→ source of obligation

Damage

→ possible consequence

Remedy

→ potentially damages, performance or other contractual relief.

But it still needs to answer questions such as:

• Was delivery actually due on that date?
• Was the delay excused?
• Was there force majeure?
• Did the claimant contribute to the delay?
• Was the loss caused by the delay?
• Was the loss proved?
• Did the contract limit liability?

These questions demonstrate why semantic representation does not equal autonomous adjudication.

8. Electronic Transactions Law

Federal Decree-Law No. 46 of 2021 on Electronic Transactions and Trust Services is an important foundation for digital legal systems.

The legislation recognises electronic documents and electronic contracting and, importantly, recognises contracts formed between automated electronic mediums.

This means the UAE legal system already accommodates situations where software participates in legally significant transactions.

However:

Recognition of automated transactions is not recognition of autonomous legal personality.

That distinction remains essential.

9. Machine-Readable Law and Digital Courts

The DIFC provides an especially advanced UAE example.

The DIFC has established a Digital Economy Court dealing with disputes involving areas including:

• digital assets;
• blockchain;
• artificial intelligence;
• fintech;
• big data;
• cloud services;
• digital payments;
• other emerging technologies.

The significance is institutional:

technology is being integrated into adjudication without eliminating the human court.

The Techteryx litigation demonstrates the practical operation of this specialised jurisdiction.

10. Case Law

Because there are still relatively few UAE decisions specifically titled “machine-readable law,” the following cases should be understood as direct and analogous authorities concerning digital systems, AI-generated legal material, automated information and technology-intensive adjudication.

DIFC decisions are not automatically binding on mainland UAE courts.

Case 1 — Klesta Eshja & Hair Creators Salon LLC v Salah Masri & Others

CFI 066/2024 — DIFC Court of First Instance

This is one of the most important recent authorities for the limits of machine-assisted legal information.

The litigation involved AI-assisted drafting of defence material. The DIFC Court subsequently dealt with applications concerning the pleadings and their deficiencies, including issues associated with AI-generated material. The court's later orders include the March and July 2026 proceedings concerning the strike-out applications.

Principle

A computer-readable or AI-generated legal proposition is not automatically reliable merely because it has been generated in a sophisticated format.

Relevance

Machine-readable legal databases require:

• source verification;
• version control;
• authority checking;
• human supervision.

11. Case 2 — Stelian Gheorghe v BSA Ahmad Bin Hezeem & Associates LLP & Jimmy Haoula

CFI 045/2025 — DIFC Court of First Instance

The proceedings concerned a dispute that was ultimately stayed in favour of arbitration. Subsequent proceedings dealt with costs and permission to appeal.

The case is relevant to semantic legal systems because it demonstrates that legal classification of a dispute—particularly jurisdiction and arbitration—remains a judicial legal question.

Principle

A computational system may identify an arbitration clause, but determining whether the dispute falls within that clause involves legal interpretation.

Thus:

Text recognition ≠ legal conclusion.

12. Case 3 — Graciela Limited v Giacobbe

[2014] DIFC CFI 027

This case involved deliberate interference with an IT system.

The DIFC Court found the defendant responsible for deliberately interfering with the claimant's IT system and awarded approximately USD 690,533 in compensatory damages. The court dealt with the technological evidence as part of ordinary legal adjudication.

Principle

Digital systems can be:

• objects of legal protection;
• sources of evidence;
• instruments of conduct;

without becoming independent legal decision-makers.

Semantic-law significance

A semantic system might identify:

IT system + interference + damage → possible civil liability.

But a human court still determines:

• attribution;
• evidence;
• causation;
• damage;
• liability.

13. Case 4 — Linux v Lizeth

[2022] DIFC SCT 237

The case concerned a Software Development Agreement between two Dubai companies. The claimant sought payment following alleged contractual breach, but the claim was dismissed.

Principle

Software can be the subject matter of legal rights and contractual obligations.

The important conceptual point is:

Software is legally relevant without becoming the legal interpreter of the contract governing it.

Semantic relevance

A machine-readable contract could identify:

• parties;
• obligations;
• payment terms;
• deadlines;
• breach provisions.

But the legal effect of a disputed provision remains a matter of legal interpretation.

14. Case 5 — ICICI Bank Limited v Bavaguthu Raghuram Shetty

[2022] DIFC CFI 034

The final judgment was issued on 17 February 2025.

The dispute involved complex financial and documentary issues. The Court ultimately dismissed the claim under one personal guarantee and allowed claims under three NMC personal guarantees in the amount of USD 106,294,108.

Earlier procedural proceedings also involved expert evidence and questions concerning the admissibility of a supplemental expert report.

Principle

Large volumes of structured information do not eliminate the need for human legal evaluation.

Semantic significance

An automated legal system might map:

Guarantee → signature → obligation → breach → liability

but it still needs legal and evidentiary analysis concerning:

• authenticity;
• contractual construction;
• applicable law;
• evidence;
• burden of proof.

15. Case 6 — Techteryx Ltd v Aria Commodities DMCC & Others

[2025] DIFC DEC 001

This is one of the most technologically significant UAE-region cases.

The dispute concerned approximately USD 456 million in reserves backing the TrueUSD stablecoin. The Digital Economy Court granted proprietary and worldwide freezing relief concerning the relevant assets.

The case demonstrates that sophisticated digital assets can be incorporated into conventional legal concepts such as:

• ownership;
• tracing;
• proprietary relief;
• freezing orders;
• disclosure.

Semantic significance

A semantic legal system could connect:

Stablecoin

→ digital asset

→ underlying reserve

→ beneficial ownership

→ proprietary claim

→ tracing

→ remedy.

But the court still determines whether those relationships legally exist.

16. Case 7 — Alarabi Investments Limited v Cron AI Ltd

CFI 030/2025 — DIFC Court of First Instance

The defendant was Cron AI Ltd, an AI-related corporate entity.

The Court's 26 June 2026 order dealt with an application concerning the setting aside of a default judgment and subsequent procedural steps.

Principle

An AI-related business can itself be a legally recognised corporate party.

This illustrates an important distinction:

AI technology and the legal entity operating AI technology are not necessarily the same legal subject.

Semantic relevance

A legal database must distinguish:

• company;
• software;
• AI model;
• developer;
• user;
• owner;
• contractual party.

Failure to distinguish these concepts could produce serious legal errors.

17. Case 8 — Digital-Economy Court in Techteryx: Continuing Proceedings

The Techteryx litigation has continued through 2026, with further Digital Economy Court orders concerning disclosure, compliance, contempt and costs.

This is significant for machine-readable legal systems because it demonstrates that legal status changes over time.

A legal database must therefore record:

• original order;
• amended order;
• subsequent order;
• current status;
• parties;
• procedural history;
• continuing obligations.

Principle

Legal information is temporal.

A machine-readable legal system that does not maintain version history can produce legally obsolete answers.

18. DIFC Practical Guidance on Generative AI

The DIFC Courts' Practical Guidance Note No. 2 of 2023 is especially important.

It identifies risks including:

• incorrect or misleading information;
• confidentiality breaches;
• intellectual-property problems;
• data-protection issues.

It also requires verification of AI-generated content and states that AI should assist rather than replace integral human decision-making in preparing submissions.

This is directly relevant to semantic legal systems.

The Court's approach can be summarised as:

Machine assistance + verification + transparency + human responsibility

rather than:

Machine output = law

19. Semantic Ambiguity

Legal language is often deliberately flexible.

Consider:

• “reasonable time”;
• “good faith”;
• “substantial loss”;
• “appropriate compensation”;
• “public order”;
• “abuse of rights”.

A machine-readable system can encode these expressions as legal concepts.

But it cannot necessarily establish a single numerical threshold.

For example:

What exactly constitutes a “reasonable time”?

The answer can depend on:

• contract;
• commercial practice;
• industry;
• circumstances;
• conduct of the parties;
• surrounding facts.

This is known as the semantic indeterminacy problem.

20. Polysemy and Legal Meaning

One legal word can have different meanings depending on context.

For example:

“Interest”

could mean:

• contractual interest;
• statutory interest;
• beneficial interest;
• economic interest;
• legal interest.

A simplistic keyword-based system may incorrectly treat them as identical.

A semantic system therefore requires:

context + jurisdiction + legal domain + temporal status

before assigning legal meaning.

21. Cross-Jurisdictional Problem

This is particularly important in the UAE.

The legal environment includes:

• Federal UAE law;
• Emirate-level laws;
• DIFC law;
• ADGM law;
• arbitration rules;
• free-zone regulations;
• international conventions.

A semantic legal engine must determine:

Which legal universe governs this particular dispute?

For example:

Dubai mainland dispute

is not automatically governed by:

DIFC law

merely because a DIFC case contains similar language.

22. Temporal Versioning

Machine-readable law requires a time dimension.

Suppose:

• Article A existed in 2024;
• it was amended in 2025;
• new legislation became effective in 2026.

The computer must answer:

Which version applied on the date of the transaction?

This is particularly important in the UAE because the new Civil Transactions Law replaced the previous general Civil Transactions Law framework with effect from 1 June 2026.

Therefore, a legal AI system must distinguish:

law at time of transaction

from

law currently in force.

23. Machine-Readable Law and Repeal

A sophisticated legal database should not simply delete repealed provisions.

Instead, it should record:

Article X

→ enacted

→ amended

→ effective date

→ repealed

→ replacement provision

→ transitional rule.

This is essential for civil disputes involving:

• old contracts;
• continuing obligations;
• limitation periods;
• accrued rights;
• transitional provisions.

24. The Problem of Legal Exceptions

Legal reasoning is heavily dependent on exceptions.

A simplistic computational rule might say:

Contract + breach = damages.

But actual legal reasoning may be:

Contract + breach + causation + proved loss + recoverability + no applicable exclusion + mitigation considerations = potential damages.

Therefore:

Exceptions are as important as rules.

A semantic legal system must model both.

25. Rules, Principles and Standards

Legal rules are easier to encode than legal standards.

Rule

A filing must be made within X days.

Easy to encode.

Principle

Parties must act in good faith.

More difficult.

Standard

Conduct must be reasonable in the circumstances.

Even more context-dependent.

Justice-based principle

Apply the rules of justice where no specific legal rule exists.

Extremely difficult to reduce to deterministic computational instructions.

This distinction is particularly important under Article 1 of the current Civil Transactions Law.

26. Semantic Legal Systems and Judicial Discretion

A semantic engine could assist a judge by presenting:

• applicable statutes;
• conflicting authorities;
• previous interpretations;
• relevant facts;
• possible remedies.

But it should not automatically convert those inputs into a binding judgment.

The reason is that judicial reasoning may involve:

1. interpretation;
2. evidentiary evaluation;
3. credibility;
4. proportionality;
5. contextual assessment;
6. legal policy;
7. justice.

Therefore:

Semantic legal technology should generally be viewed as decision-support infrastructure rather than an autonomous source of judicial authority.

27. Machine-Readable Contracts

The concept can also be applied to contracts.

A machine-readable contract can identify:

• party;
• obligation;
• deadline;
• payment;
• condition;
• termination event;
• penalty;
• dispute-resolution mechanism.

A semantic contract can go further:

Payment obligation → triggered by → delivery → verified by → acceptance certificate.

This could enable automated compliance monitoring.

But disputes can still arise concerning:

• whether delivery occurred;
• whether acceptance was valid;
• whether force majeure applies;
• whether the party waived a right;
• whether the contract was varied.

Thus:

machine-readable contract ≠ machine-determined contract meaning.

28. Smart Contracts and Semantic Law

Smart contracts provide an important example.

Traditional contract:

Legal text → human interpretation → performance

Smart contract:

Legal agreement → code → automated performance

The problem arises when:

legal meaning ≠ coded instruction.

For example, code may automatically transfer an asset even though a legal dispute exists concerning:

• fraud;
• mistake;
• invalid consent;
• authority;
• contractual interpretation.

The legal system may therefore need to distinguish:

what the code executed

from

what the law required.

29. Machine-Readable Law and Evidence

Semantic legal systems also create evidentiary questions.

A court may ask:

Source

Where did the machine obtain the rule?

Version

Which version of the statute did it use?

Metadata

Was the provision properly tagged?

Transformation

Was human-readable text converted correctly?

Algorithm

How did the system select the relevant provision?

Reproducibility

Would the same input generate the same result?

Auditability

Can the result be reconstructed later?

These questions become increasingly important as legal systems become computational.

30. Audit Trails

A reliable semantic legal system should maintain an audit trail showing:

1. input;
2. legal database version;
3. statute version;
4. cases consulted;
5. algorithm/model used;
6. assumptions;
7. output;
8. human modifications.

This allows a court or regulator to reconstruct how a legal conclusion was produced.

Without an audit trail:

machine reasoning can become difficult to challenge.

31. Semantic Bias

Machine-readable systems can create a new form of legal bias.

Suppose the database:

• contains more English-language cases;
• emphasises recent decisions;
• underrepresents particular legal concepts;
• incorrectly ranks authorities;
• treats frequent outcomes as more important.

The resulting system could unintentionally distort legal research.

Therefore:

Data architecture can influence legal reasoning even without deliberately changing the law.

32. Explainability

A semantic legal system should ideally be able to answer:

“Why did you select this provision?”

and:

“Why did you exclude this exception?”

and:

“Which version of the law did you use?”

and:

“Which cases support this interpretation?”

This is much more useful legally than simply producing:

“Answer: Claimant succeeds.”

33. Human-in-the-Loop Model

A suitable model for UAE legal technology is:

Human facts

↓

Machine legal retrieval

↓

Semantic classification

↓

Potential legal rules

↓

Human verification

↓

Legal interpretation

↓

Human decision

↓

Reasoned judgment

This preserves technological efficiency while retaining human legal responsibility.

34. Five Levels of Automation

The UAE's present institutional development is substantially focused on Levels 2–4, rather than granting autonomous legal decision-making authority to machines.

The DIFC's AI guidance expressly emphasises assistance rather than replacement of integral human decision-making.

35. Important Legal Limits

A. Definitive statutory text

A machine cannot override a definitive legislative provision.

B. Jurisdiction

The system must identify the correct court and legal regime.

C. Time

The correct version of the law must be used.

D. Exceptions

Exceptions must be recognised rather than ignored.

E. Evidence

Machine-generated conclusions do not automatically establish facts.

F. Interpretation

Legal ambiguity requires contextual interpretation.

G. Public order

Certain legal principles cannot simply be reduced to commercial preferences.

H. Human responsibility

The person submitting or relying upon AI-generated legal material remains accountable.

36. Mainland UAE, DIFC and ADGM

37. Six Core Case-Law Lessons

38. Key Doctrinal Formulas

For examination purposes, remember:

Formula 1

Machine-readable law ≠ machine-made law

Formula 2

Semantic classification ≠ legal interpretation

Formula 3

Legal prediction ≠ legal judgment

Formula 4

Automated execution ≠ legal personality

Formula 5

Digital evidence ≠ automatically proven evidence

Formula 6

Structured legislation ≠ completely deterministic law

39. Major Future Issues

The UAE's continued development of digital legal infrastructure is likely to raise questions concerning:

1. standardised legislative data formats;
2. machine-readable amendments;
3. legal ontologies;
4. AI-assisted statutory interpretation;
5. automated compliance;
6. smart contracts;
7. digital evidence;
8. algorithmic judicial assistance;
9. explainability;
10. auditability;
11. legal-data provenance;
12. cross-border semantic interoperability;
13. machine-readable arbitration clauses;
14. automated enforcement;
15. AI-generated legal submissions.

40. Exam-Oriented Short Note

Machine-readable law is the conversion of legislation and legal information into structured data that computers can process.

Semantic legal systems go further by representing relationships between legal concepts, rules, exceptions, facts and remedies.

In UAE civil law, this technology is particularly relevant because the legal environment increasingly incorporates electronic transactions, digital evidence and specialised digital adjudication. The new Civil Transactions Law nevertheless preserves a hierarchy of legal sources and interpretive principles that cannot simply be replaced by algorithmic outputs.

The DIFC's Practical Guidance Note No. 2 of 2023 is particularly important because it requires verification of AI-generated material, transparency regarding AI use, protection of confidentiality and avoidance of excessive reliance on generative AI.

Cases such as Klesta Eshja, Graciela, Linux v Lizeth, ICICI Bank v Shetty, Techteryx, Alarabi Investments v Cron AI, and Stelian Gheorghe demonstrate different aspects of the relationship between technology and legal adjudication.

41. Conclusion

The development of machine-readable law represents an important stage in the digital transformation of UAE civil justice.

Its greatest benefit is structured access to law:

statutes → concepts → relationships → cases → evidence → potential legal rules.

But the major limitation is that law is not merely a database of rules.

Legal reasoning can involve:

• ambiguity;
• context;
• competing principles;
• evidence;
• judicial discretion;
• public order;
• custom;
• justice;
• changing factual circumstances.

The current UAE approach therefore supports a human-supervised semantic legal system rather than a system in which algorithms independently determine the law.

The most important examination proposition is:

Machine-readable law can make UAE law more searchable, structured, interconnected and computationally usable; semantic legal systems can assist interpretation by mapping legal concepts and relationships, but neither technology automatically acquires the authority to determine the legal meaning of every rule or resolve every civil dispute.

In short: Law can be made machine-readable; legal judgment cannot simply be reduced to machine-readable rules.