Civil Law and UAE Autonomous Arbitration Engines in Global Commerce

1. Introduction

Autonomous arbitration engines are AI-enabled or highly automated systems designed to perform functions traditionally undertaken by arbitral institutions, tribunals, case managers, or legal professionals.

In global commerce, such systems may be used to:

identify the applicable arbitration clause;

determine whether a dispute falls within a contract;

appoint or recommend arbitrators;

organize evidence;

generate procedural timetables;

conduct document review;

analyze contractual obligations;

assess damages;

propose settlements;

generate draft procedural orders;

assist with award drafting; and

potentially generate automated dispute outcomes.

The most important legal question is:

Can an autonomous technological system itself exercise the adjudicative authority of an arbitral tribunal under UAE law?

The present answer should be approached cautiously. AI can assist arbitration extensively, but an autonomous engine cannot simply acquire arbitral jurisdiction merely because the parties have programmed it into a contract. The authority of an arbitral tribunal ultimately derives from a valid arbitration agreement and applicable arbitration law.

For UAE-seated arbitration, the principal statute remains Federal Law No. 6 of 2018 on Arbitration. The UAE's current civil-law framework must also be considered, particularly the new Civil Transactions Law effective 1 June 2026.

2. Meaning of an Autonomous Arbitration Engine

An autonomous arbitration engine can be described as:

A software or AI system capable of performing some or all procedural, analytical, or decision-making functions associated with arbitration with limited or no continuous human intervention.

There are several levels of autonomy.

Level 1 — Administrative automation

The system handles:

filing;

scheduling;

notifications;

document organization;

deadlines.

This is relatively low risk.

Level 2 — Analytical assistance

AI analyzes:

contracts;

evidence;

correspondence;

damages;

chronology;

legal authorities.

This is more sophisticated but still primarily assistive.

Level 3 — Decision support

The system recommends:

jurisdictional outcomes;

procedural orders;

liability;

damages;

costs.

Human arbitrators remain decision-makers.

Level 4 — Autonomous adjudication

The system itself decides:

jurisdiction;

liability;

damages;

costs; and

final disposition.

This raises serious questions concerning consent, impartiality, due process, arbitrator appointment, legal personality, accountability, challenge and enforcement.

3. Fundamental UAE Principle: Arbitration Comes From Consent

The UAE Arbitration Law recognizes arbitration as a consensual dispute-resolution mechanism.

Under Articles 4–7 of Federal Law No. 6 of 2018, the arbitration agreement must satisfy requirements relating to:

capacity;

authority;

writing;

consent;

scope;

arbitrability.

This creates the first difficulty for an autonomous arbitration engine.

A contract saying:

"All disputes shall be decided by an AI arbitration engine"

does not necessarily answer all legal questions.

The system must still have a legally valid basis for exercising adjudicative authority.

4. The Arbitration Agreement and AI

An arbitration agreement may exist:

as an arbitration clause;

in a separate agreement;

after the dispute arises;

through qualifying electronic communications;

through incorporation by reference.

An autonomous engine cannot replace the arbitration agreement.

The legal sequence remains:

Valid arbitration agreement

↓

Dispute within its scope

↓

Properly constituted tribunal/system

↓

Procedurally valid arbitration

↓

Award

↓

Recognition/enforcement

The AI cannot reverse this hierarchy.

5. Article 6 and Separability

The UAE Arbitration Law recognizes the principle of separability.

The arbitration clause is legally distinct from the underlying contract.

This is highly relevant to autonomous arbitration.

Suppose an AI system automatically executes a smart contract and a party alleges:

"The underlying contract is invalid."

The arbitration clause may nevertheless survive for the purpose of determining the dispute concerning the contract.

However, this does not mean that an AI engine automatically possesses jurisdiction.

The validity of the arbitration agreement itself must still be established.

6. Competence-Competence

Under the UAE Arbitration Law, the arbitral tribunal has authority to determine questions concerning its jurisdiction, including the existence and validity of the arbitration agreement.

This is known as competence-competence.

An autonomous arbitration engine therefore faces a fundamental question:

Can software determine its own jurisdiction?

Technically, an AI system can perform jurisdictional analysis.

Legally, however, the authority to decide jurisdiction must derive from the arbitration agreement and applicable arbitration legislation.

Therefore:

AI capability ≠ arbitral jurisdiction.

7. Autonomous Arbitration and the Human Arbitrator

Traditional arbitration assumes an arbitrator or tribunal capable of:

exercising judgment;

assessing evidence;

maintaining independence;

maintaining impartiality;

observing procedural fairness;

giving reasons;

determining liability; and

issuing an award.

An autonomous AI system does not naturally fit the traditional concept of an arbitrator.

Important questions include:

Is AI capable of being appointed as an arbitrator?

Who is responsible for the system?

Who bears liability for an erroneous decision?

Can the AI be challenged for bias?

Can the parties cross-examine the algorithm?

Who gives the oath or undertaking of impartiality?

Who signs the award?

Can the system be removed?

Who corrects an algorithmic error?

These questions make fully autonomous arbitration legally uncertain.

8. Autonomous Arbitration and Due Process

Due process is one of the most important safeguards.

The parties should have an adequate opportunity to:

present their case;

respond to the opposing case;

produce evidence;

challenge evidence;

make legal submissions;

participate in hearings where appropriate.

An autonomous system must therefore not silently use information that one party cannot inspect or challenge.

For example:

AI system receives undisclosed external data → incorporates data into liability determination → issues award.

This could create a serious procedural-fairness problem.

9. Equality of Treatment

The arbitral process must treat parties fairly.

AI systems can potentially introduce unequal treatment through:

different datasets;

different language processing;

biased training data;

different confidence thresholds;

inconsistent interpretation of similar clauses.

Therefore, an autonomous arbitration engine should apply:

same legal framework + same procedural standards + transparent methodology

to both parties.

10. AI Bias and Arbitrator Impartiality

Traditional arbitrator bias may involve:

financial interests;

relationships;

prior involvement;

institutional conflicts.

AI introduces different forms of bias:

training-data bias;

model bias;

algorithmic weighting;

language bias;

historical case bias;

developer assumptions.

The parties therefore need transparency concerning:

model architecture;

data sources;

system version;

material updates;

decision criteria;

conflicts involving developers or providers.

11. UAE Case 1: Dubai Court of Cassation, Case No. 735/2024

This is an important recent authority concerning the interpretation and validity of arbitration agreements.

The Court addressed statutory requirements relating to arbitration and the parties' intention to arbitrate.

Relevance to autonomous arbitration

The case supports a fundamental proposition:

Arbitration depends upon legally recognizable consent.

Therefore, simply inserting an AI arbitration mechanism into software does not automatically establish valid arbitration.

A contract should clearly establish:

agreement to arbitrate;

scope;

applicable rules;

appointment mechanism;

seat;

applicable law;

procedural framework.

12. UAE Case 2: Abu Dhabi Court of Cassation, Case No. 902/2024

This case concerned an alleged defect involving the authority of the person signing the arbitration agreement.

Principle

The authority of the person agreeing to arbitration is legally significant.

Relevance to autonomous arbitration

This is particularly important for smart contracts.

Suppose an employee deploys a blockchain contract containing an AI arbitration clause.

The question remains:

Did the employee have authority to bind the company to arbitration?

Blockchain immutability cannot cure a lack of corporate authority.

Thus:

technical execution ≠ legal authority.

13. UAE Case 3: Dubai Court of Cassation, Case No. 756/2024

This case addressed the seat of arbitration and the distinction between the legal seat and the physical location where hearings occur.

Relevance

An autonomous arbitration engine operating globally could theoretically have:

software hosted in one country;

parties in several countries;

hearings conducted online;

data stored in another country.

The system must nevertheless identify the legal seat.

The seat determines important legal consequences, including the supervisory court and applicable arbitration framework.

Therefore:

Server location is not necessarily the arbitral seat.

14. UAE Case 4: UAE Federal Supreme Court, Case No. 873/JY3 of 2009

This earlier UAE authority recognized the contractual nature of an arbitration agreement.

Relevance

Although it predates the 2018 Arbitration Law, the case is useful as a foundational authority for understanding arbitration as a matter of contractual consent.

An autonomous arbitration engine cannot independently create an arbitration relationship where the parties never validly agreed to arbitrate.

15. UAE Case 5: Dubai Court of Cassation, Case No. 33/2009

This earlier case concerned contractual arbitration.

Relevance

It supports the broader principle that arbitration must be connected to a valid contractual agreement.

For autonomous systems, this means:

The code may execute the arbitration procedure, but the legal authority originates in the parties' agreement.

The case is historical/foundational because it predates the current UAE Arbitration Law.

16. UAE Case 6: Dubai Court of Cassation, Case No. 220/2004

This authority concerned arbitration clauses incorporated through contractual arrangements and general conditions.

Relevance to automated commerce

Global digital commerce frequently operates through:

standard terms;

platform terms;

APIs;

electronic purchase orders;

master agreements;

incorporated schedules.

An autonomous arbitration engine should therefore verify whether an arbitration clause was properly incorporated, rather than assuming that a reference to another document automatically incorporates its arbitration provision.

17. UAE Case 7: UAE Cassation No. 275/2010, Hearing of 1 June 2011

This authority emphasized the importance of special authority to agree to arbitration.

The reasoning is significant because arbitration involves a waiver of ordinary court adjudication.

Relevance to autonomous arbitration

A company's ordinary employee authority to enter commercial transactions does not necessarily establish authority to agree to arbitration.

Therefore an automated contracting system should distinguish:

authority to contract

from

authority to agree to arbitration.

18. UAE Case 8: Ginette PJSC v Geary Middle East FZE & Geary Limited, DIFC Court of Appeal, CA-005-2016

The DIFC Court of Appeal considered issues involving arbitration clauses, incorporation and the authority required to agree to arbitration under UAE law.

Relevance

The case demonstrates the importance of:

clear incorporation;

contractual consent;

authority;

arbitration-specific requirements.

It is particularly useful for digital contracting because global commercial platforms often incorporate arbitration provisions through linked terms or contractual documents.

19. UAE Case 9: DIFC CFI ARB-018/2023

This DIFC authority concerned the validity and incorporation of arbitration clauses under UAE law.

The Court considered circumstances in which contractual language incorporated arbitration provisions by reference.

Relevance to autonomous arbitration

An AI system should not assume that an arbitration clause contained somewhere within a digital document automatically binds every participant.

It should identify:

the incorporated document;

the incorporation language;

whether the arbitration provision was clearly incorporated;

whether the relevant party accepted it.

20. UAE Case 10: Abu Dhabi Court of Cassation, Case No. 922/2020

This case is relevant to contractual authority and good-faith performance of arbitration-related obligations.

Relevance

An autonomous arbitration system must operate within the contractual framework agreed by the parties.

It should not:

expand the scope of arbitration;

alter the parties;

change the seat;

modify procedural rights;

impose new obligations

without a valid legal basis.

21. Autonomous Arbitration and Non-Signatories

Global commerce frequently involves:

parent companies;

subsidiaries;

agents;

subcontractors;

platform operators;

guarantors;

affiliates.

An autonomous engine might identify a relationship and conclude:

"Company B is commercially connected to Company A, therefore Company B is bound by the arbitration clause."

That is legally dangerous.

Commercial affiliation alone does not automatically establish consent to arbitration.

A valid legal basis must exist for extending the arbitration agreement to a non-signatory.

22. Autonomous Appointment of Arbitrators

An AI system could potentially recommend an arbitrator based upon:

expertise;

language;

availability;

nationality;

prior experience;

conflicts.

This can be useful.

But appointment must comply with:

arbitration agreement;

institutional rules;

statutory requirements;

independence;

impartiality;

challenge procedures.

An algorithm should therefore recommend or facilitate appointment, unless a valid legal framework expressly permits the relevant automated mechanism.

23. Autonomous Tribunal Versus AI-Assisted Tribunal

This distinction is essential.

AI-assisted tribunal

Human arbitrators + AI tools

The arbitrators remain responsible for:

evidence;

reasoning;

findings;

award.

This is considerably easier to reconcile with existing arbitration law.

Autonomous tribunal

AI system + no meaningful human decision-maker

This raises unresolved questions about:

arbitrator status;

impartiality;

capacity;

accountability;

due process;

enforceability.

Therefore, the second model presents substantially greater legal uncertainty.

24. AI and the Arbitral Award

A final award should normally identify:

tribunal;

parties;

jurisdiction;

issues;

reasoning;

findings;

relief;

costs;

date and place/seat as required.

If an AI system generates an award, important questions arise:

Who is legally the author?

Who signs it?

Who accepts responsibility?

Was the tribunal properly constituted?

Were the parties heard?

Were reasons actually determined by the tribunal?

Can the award be challenged?

A system-generated document should therefore not automatically be treated as a legally valid arbitral award.

25. Enforcement Problem

This is perhaps the greatest difficulty.

An award may ultimately need recognition or enforcement in:

UAE;

Saudi Arabia;

Singapore;

India;

United Kingdom;

United States;

European jurisdictions;

other New York Convention states.

The enforcing court may ask whether:

a valid arbitration agreement existed;

the tribunal was properly constituted;

parties received proper notice;

parties had an opportunity to present their case;

the award exceeded the arbitration agreement;

the award violates public policy;

procedural requirements were satisfied.

An autonomous engine does not escape these requirements.

26. New York Convention Considerations

Global commerce frequently relies on the 1958 New York Convention for recognition and enforcement of foreign arbitral awards.

The Convention's framework makes the validity of the arbitration agreement and procedural fairness extremely important.

An AI-generated award therefore faces the same enforcement questions as a conventional award.

The technological novelty of the tribunal does not automatically remove the enforcing court's supervisory role.

27. Autonomous Arbitration and Public Policy

Public policy creates another limitation.

Suppose an AI system produces an award based on an algorithm that:

discriminates against one party;

relies upon secret evidence;

denies an opportunity to respond;

uses fabricated evidence;

applies an impermissible rule;

exceeds the tribunal's jurisdiction.

The award could face serious enforcement problems.

Thus:

Algorithmic efficiency cannot override public policy.

28. Confidentiality

Arbitration is frequently commercially confidential.

AI systems may process:

trade secrets;

pricing;

customer lists;

intellectual property;

financial records;

technical information.

Sending these materials into an external AI model can create:

confidentiality risks;

cybersecurity risks;

unauthorized disclosure;

data-transfer issues;

privilege issues.

Therefore, an autonomous arbitration engine should use strong controls over:

data access;

storage;

model training;

retention;

deletion;

third-party access.

29. AI and Evidence in Arbitration

AI can be particularly useful for:

document review;

email classification;

chronology;

financial analysis;

translation;

transcript preparation;

metadata analysis.

But the tribunal should know whether material was:

human-generated

or

AI-generated.

For example:

AI-generated translation → must be checked where accuracy is material.

AI-generated summary → should not automatically replace the underlying evidence.

AI-generated forensic conclusion → may require expert explanation.

30. Hallucination Risk

One of the greatest dangers is AI hallucination.

An autonomous arbitration engine could potentially:

invent case law;

invent contractual provisions;

misinterpret legislation;

misstate evidence;

fabricate factual connections.

This is particularly dangerous because an arbitral award must be based on the actual record.

A robust system therefore requires:

retrieval + source verification + evidence anchoring + human validation.

31. Autonomous Arbitration and Natural Justice

A legally defensible autonomous system would need mechanisms for:

Notice

The party knows what dispute is being decided.

Opportunity to respond

The party can challenge the opposing case.

Evidence access

Relevant evidence is available subject to legitimate confidentiality restrictions.

Equality

Both parties have equivalent procedural opportunities.

Reasoned decision

The basis of the outcome can be understood.

Review/challenge

Applicable legal mechanisms remain available.

These safeguards are difficult to guarantee with a completely opaque algorithm.

32. Automated Damages Calculation

AI can calculate:

unpaid invoices;

interest;

contractual penalties;

lost profits;

cost escalation;

delay damages.

However, damages involve legal concepts such as:

causation;

proof;

contractual limitations;

mitigation;

foreseeability;

actual loss.

Therefore:

AI calculation ≠ legal determination of damages.

The system can calculate numbers, but the tribunal must determine which numbers are legally recoverable.

33. Autonomous Arbitration in Smart Contracts

Consider a blockchain-based international supply contract:

Oracle confirms delivery → smart contract detects dispute → AI engine evaluates evidence → AI decides seller breached → cryptocurrency automatically transferred to buyer.

This appears highly efficient.

But legal questions remain:

Was the arbitration agreement valid?

Was the AI properly authorized?

Was the oracle accurate?

Was the evidence authentic?

Was the decision procedurally fair?

Was the award valid?

Can the automatic transfer be reversed?

Was the remedy within the tribunal's jurisdiction?

Therefore, blockchain execution should not be confused with legally final adjudication.

34. Autonomous Arbitration and Emergency Relief

AI could help identify urgent circumstances involving:

dissipation of assets;

destruction of evidence;

intellectual-property misuse;

confidentiality breaches;

threatened transfer of property.

However, emergency relief requires legal authority.

The system should therefore distinguish:

risk detected

from

legally authorized interim measure.

35. Seat of Arbitration and Digital Arbitration

Digital arbitration creates an interesting distinction.

A case may involve:

parties in Dubai;

tribunal members in London and Singapore;

hearing conducted online;

AI software hosted in the United States;

documents stored in cloud infrastructure in Europe.

None of these facts necessarily determines the legal seat.

The parties' arbitration agreement and applicable arbitration law determine the legal framework.

The principle discussed in Dubai Court of Cassation Case No. 756/2024 is therefore especially important.

36. Autonomous Arbitration and Institutional Rules

Institutional arbitration may provide additional safeguards.

An AI engine could operate within an institutional framework as:

document-management system;

procedural assistant;

scheduling tool;

legal research tool;

transcription tool;

analytical assistant.

The institution and tribunal would retain ultimate responsibility.

This model is much easier to reconcile with existing arbitration structures than completely autonomous adjudication.

37. Recommended UAE Model

A practical model for UAE global commerce would be:

Stage 1

AI dispute detection

↓

Stage 2

Human confirmation that dispute exists

↓

Stage 3

Verification of arbitration agreement

↓

Stage 4

Jurisdictional assessment

↓

Stage 5

Human arbitrator/tribunal constitution

↓

Stage 6

AI-assisted evidence analysis

↓

Stage 7

Human hearing and submissions

↓

Stage 8

AI-assisted drafting/analysis

↓

Stage 9

Human tribunal independently determines award

↓

Stage 10

Human-authenticated award

↓

Stage 11

Recognition/enforcement

This approach captures the efficiency of AI while preserving legal accountability.

38. Legal Risks of Fully Autonomous Arbitration

39. Autonomous Arbitration and Civil-Law Principles

The relationship can be summarized as follows:

40. The Most Important Legal Distinction

The most important distinction is:

Autonomous arbitration assistance

versus

Autonomous arbitration adjudication

AI assistance is already capable of providing substantial value.

For example:

AI reviews 500,000 documents in two hours and identifies potentially relevant evidence.

There is no conceptual requirement that the AI itself be the arbitrator.

But:

AI independently decides that Party A owes Party B USD 50 million.

This raises substantially more fundamental questions about arbitral authority and enforceability.

41. Can Parties Agree to AI Arbitration?

Parties can potentially agree contractually to sophisticated automated dispute-resolution mechanisms.

However, the agreement should clearly address:

whether AI is merely advisory;

whether a human arbitrator remains involved;

appointment;

procedural rules;

applicable law;

seat;

confidentiality;

data security;

model transparency;

evidence;

challenge mechanisms;

correction of algorithmic errors;

award authentication;

enforcement.

A vague clause saying:

"All disputes shall be decided by AI"

is substantially more problematic than a detailed clause providing an AI-assisted arbitration process under a recognized arbitration framework.

42. Recommended Contractual Clause Structure

For global commerce, an AI-assisted arbitration clause should ideally identify:

Seat

Governing law

Institution/rules

Number of arbitrators

Appointment procedure

Permitted AI use

Human decision-maker

Confidentiality

Data protection

Evidence requirements

Challenge procedures

Award authentication

Emergency relief

Enforcement mechanism

This reduces uncertainty.

43. Practical Example

Imagine a UAE company enters into a global commodities contract with a Singapore company.

The contract provides:

"Disputes shall be resolved by arbitration seated in Dubai."

A dispute arises concerning delivery.

The arbitration platform automatically:

identifies the arbitration clause;

collects invoices;

analyzes shipping records;

compares delivery dates;

calculates claimed damages;

identifies relevant correspondence;

produces a draft chronology.

A human arbitral tribunal then:

confirms jurisdiction;

hears both parties;

evaluates evidence;

determines liability;

determines damages;

issues the final award.

This is a highly practical form of AI-assisted arbitration.

44. Fully Autonomous Version

Now consider:

AI system receives claim → AI decides jurisdiction → AI evaluates evidence → AI determines liability → AI calculates damages → blockchain automatically transfers funds.

This model presents significantly greater legal uncertainty.

The primary issues are:

tribunal constitution;

consent;

impartiality;

due process;

accountability;

reasoned award;

challenge;

enforcement.

Until these issues are clearly resolved by legislation, institutional rules and judicial decisions, human-supervised arbitration is the safer UAE model.

45. Conclusion

Autonomous arbitration engines in UAE global commerce have substantial potential to transform international dispute resolution.

They can dramatically improve:

speed;

document review;

procedural administration;

evidence organization;

damages calculation;

legal research;

multilingual arbitration;

compliance;

cost management.

However, the fundamental UAE arbitration principles remain critical.

The most important rule is:

Arbitral authority comes from a valid arbitration agreement and applicable arbitration law, not from the technological sophistication of an AI system.

The UAE authorities discussed above—particularly Dubai Court of Cassation Cases 735/2024 and 756/2024, Abu Dhabi Court of Cassation Case 902/2024, UAE Cassation No. 275/2010, Ginette PJSC v Geary, and DIFC ARB-018/2023—illustrate the continuing importance of consent, authority, incorporation, jurisdiction and the legal seat.

Accordingly, the most defensible model for UAE global commerce is:

Valid arbitration agreement → legally constituted human tribunal → AI-assisted evidence and analysis → human adjudication → reasoned award → judicial recognition/enforcement.

A completely autonomous AI arbitrator that independently exercises adjudicative power remains a much more difficult legal proposition. AI may automate the arbitration process, but it should not be assumed to possess arbitral authority merely because the parties have embedded it in software or a smart contract.

The older cases cited above are primarily foundational/analogical authorities, because most pre-date the current 2026 legal environment. The current analysis should be read together with Federal Law No. 6 of 2018 on Arbitration, the current UAE Evidence Law, and the Civil Transactions Law effective from 1 June 2026.