Civil Law and UAE: Multi-Layer Causation Mapping Systems

1. Introduction

Multi-layer causation mapping is a method of analysing civil liability where damage results from a sequence of interconnected events rather than from one isolated act.

It is especially important in modern UAE disputes involving:

• artificial intelligence;
• automated decision-making;
• multi-agent systems;
• fintech;
• cybersecurity;
• digital platforms;
• construction projects;
• medical systems;
• supply chains;
• financial transactions;
• autonomous machinery;
• data processing.

The central question is:

When several acts, omissions, systems, people and external events combine to produce damage, which of those layers legally caused the damage?

Under the current UAE Civil Transactions Law, Federal Decree-Law No. 25 of 2025, a harmful act gives rise to compensation, and the law expressly distinguishes between direct harm and harm occurring through causation. Article 247 provides that harm may occur directly or through causation and establishes rules where a direct actor and causal actor combine.

2. Meaning of Multi-Layer Causation

Traditional causation can be represented as:

Act → Damage

Multi-layer causation is more complicated:

Layer 1 → Layer 2 → Layer 3 → Layer 4 → Damage

For example:

Defective software → incorrect AI recommendation → human approval → automated transaction → financial loss

There may be several legally relevant questions at every layer.

Example

A UAE bank uses an AI fraud-detection system.

1. The data provider supplies inaccurate data.
2. The AI system interprets the data incorrectly.
3. The risk-management system produces a false risk score.
4. A bank employee relies on the score.
5. The payment system automatically blocks a legitimate transaction.
6. The customer loses a major commercial opportunity.

The customer's loss does not have one obvious cause.

The court must map the causal chain.

3. Current UAE Legal Foundation

The current Civil Transactions Law contains the basic framework.

Article 245 — Persons Subject to Civil Liability

The harmful-act provisions apply to liability arising from harmful acts committed by a natural or legal person, subject to special legislation.

This is important in technologically complex disputes because the legal system generally looks for a responsible natural or juridical person rather than treating every technological process as an independent legal person.

4. Article 246 — General Liability Rule

Article 246 provides that every act causing harm to another obliges its perpetrator to compensate for the damage, even where the perpetrator lacks discernment.

The basic structure is therefore:

Act → Harm → Responsibility → Compensation

For multi-layer causation, however, the court must go further and identify the legally significant connection between the different layers.

5. Article 247 — Direct Causation and Causation

Article 247 is particularly important.

It expressly provides that:

• harm may be direct;
• harm may occur by causation;
• direct harm attracts liability without the additional conditions applicable to consequential causation;
• where direct and causal actors combine, the legal ruling is attributed to the direct actor. 

This creates a useful starting point for multi-layer causation mapping.

Simplified model

Direct actor

↓

Causal contributor

↓

Intermediate event

↓

Damage

The court must determine where the legally relevant responsibility attaches.

6. Article 253 — Multiple Responsible Persons

Where several persons contribute to the same harm, Article 253 becomes particularly important.

The provision allows the court to address responsibility where more than one person is responsible and provides a framework for joint and several responsibility and allocation of responsibility.

This is crucial for:

• multiple contractors;
• multiple professionals;
• AI developers and operators;
• data providers;
• banks and payment providers;
• manufacturers and distributors;
• several negligent defendants.

Thus:

Multi-layer causation does not necessarily require identifying only one responsible person.

Several persons can potentially participate in the legally relevant causal chain.

7. Article 255 — Natural Consequence and Compensation

Article 255 links compensation to the loss suffered and lost profit where the loss constitutes the natural consequence of the harmful act.

This prevents unlimited claims for every event occurring after the defendant's conduct.

For example:

Software error → transaction failure → loss of customer → loss of reputation → ten years of reduced revenue

A claimant cannot automatically recover every later economic consequence.

The court must examine which losses are sufficiently connected to the wrongful conduct.

8. Multi-Layer Causation Mapping Model

A useful UAE civil-law model is:

Layer 1 — Primary Conduct

What did the defendant do or fail to do?

Layer 2 — Immediate Effect

What happened immediately afterward?

Layer 3 — Intervening Event

Did another event intervene?

Layer 4 — Human or Technological Decision

Did another person or automated system make a decision?

Layer 5 — Final Damage

What actual damage occurred?

Layer 6 — Legal Consequence

Is that damage legally attributable to the defendant?

9. Example: AI Medical System

Consider:

Incorrect medical data

↓

AI diagnostic recommendation

↓

Doctor relies on recommendation

↓

Incorrect treatment

↓

Patient injury

↓

Permanent disability

This creates several potential causal layers:

The existence of an AI system does not automatically answer the causation question.

10. Case Law

There is currently no reported UAE Supreme Court or Court of Cassation doctrine specifically called “multi-layer causation mapping.” The concept must therefore be constructed from established UAE causation principles and relevant DIFC decisions applying UAE law.

The following cases are particularly useful.

Case 1 — Larmag Holding B.V. v First Abu Dhabi Bank PJSC & Others [2019] DIFC CFI 054

This is one of the most useful authorities for UAE-law causation.

The DIFC Court examined Articles 282 and 283 of the former UAE Civil Code, which dealt with harmful acts and direct/consequential causation.

The Court recognised that UAE tort liability requires:

• harmful conduct;
• damage;
• a causal relationship between the conduct and damage.

The Court specifically discussed Article 283 as the provision concerned with causation.

Importance

The case demonstrates that causation is not merely a factual question.

The claimant must establish the legally relevant connection between:

wrongful conduct → damage

Multi-layer application

In a technological dispute, this principle can be expanded to:

Developer conduct → system behaviour → intermediate decision → damage

The claimant must establish the relevant causal bridge.

11. Case 2 — BAM Higgs & Hill LLC v Affan Innovative Structures LLC [2021] DIFC CFI 106

This is an important UAE-law damages and causation authority.

The Court expressly discussed Dubai Court of Cassation Case No. 33 of 2019 and stated that contractual or tortious liability requires the coexistence of:

1. breach/fault;
2. damage;
3. causal connection.

The Court rejected claims where alleged breaches could not properly be connected to the losses claimed.

The judgment explained that merely identifying a breach is insufficient.

Importance

This is highly relevant to multi-layer causation.

A claimant cannot simply say:

“The defendant breached the contract, therefore all subsequent losses are recoverable.”

The claimant must map the causal relationship between each breach and each claimed loss.

Formula

Breach ≠ automatic compensation

Instead:

Breach + Damage + Causation = Liability

12. Case 3 — Dubai Commercial Appeal No. 445/2020/1034

This authority was discussed in BAM Higgs & Hill.

The court stated that contractual or tortious liability depends upon:

• fault;
• damage;
• causal connection.

The causal connection may be challenged through matters such as:

• force majeure;
• external cause;
• claimant's fault;
• third-party conduct.

BAM Higgs & Hill treated the causal connection as an essential element of liability.

Multi-layer significance

This is particularly useful where the causal chain contains an intervening actor.

For example:

Defendant's error → third-party action → damage

The court must determine whether the third-party action:

• merely contributed to the loss; or
• broke the causal chain.

13. Case 4 — Dubai Court of Cassation Cases Nos. 46 and 49 of 2006 (Commercial)

These authorities were discussed in Globemed Gulf Healthcare Solutions LLC v Oman Insurance Company PSC [2017] DIFC CFI 051.

The Dubai Court of Cassation stated that compensation requires established damage or damage certain to occur, and that lost earnings must be a consequential result of the wrongful act. The claimant bears the burden of proving the lost earnings.

Importance

This establishes two important concepts:

First

There must be actual or sufficiently certain damage.

Second

Economic consequences must be causally connected to the wrongful act.

Thus:

Wrongful act → alleged lost profit

is insufficient without proof of the intermediate causal relationship.

14. Case 5 — IDBI Bank Ltd v Amira C Foods International DMCC [2019] DIFC CA 014

The DIFC Court of Appeal examined causation and loss in the context of a commercial banking dispute.

The Court emphasised that a benefit or mitigation effect should be brought into account only where it has a causal connection with the breach.

It also examined whether particular trading losses were sufficiently connected to the defendant's breach.

Importance

This demonstrates that causation operates in both directions:

Wrongful conduct → loss

and

Mitigation/benefit → reduction of loss

Both require causal analysis.

Multi-layer significance

In an AI dispute:

AI error → transaction loss → mitigation action → recovered amount

The court may need to map every stage before calculating the final compensation.

15. Case 6 — Alawwal Capital JSC v Rasmala Investment Bank Ltd, CFI 038/2023

This case provides an especially useful modern causation formulation.

The DIFC Law of Obligations required the claimant to establish that:

• but for the defendant's conduct, the claimant would not have suffered the loss; and
• the defendant's conduct was a substantial cause of the loss.

The Court applied that framework to alleged misleading representations and investment losses.

Importance

This illustrates a two-stage causation inquiry:

1. But-for causation

Would the damage have happened without the defendant's conduct?

2. Substantial causation

Was the defendant's conduct sufficiently significant to constitute a legal cause of the loss?

Multi-layer application

Suppose:

AI recommendation → human investment decision → market collapse → financial loss

The defendant may argue:

“Even without our AI recommendation, the market would have collapsed.”

That becomes a classic but-for causation issue.

16. Case 7 — Globemed Gulf Healthcare Solutions LLC v Oman Insurance Company PSC [2017] DIFC CFI 051

This case discussed Dubai Court of Cassation authorities on damage, certainty and loss of earnings.

The Court emphasised that the possibility of injury alone does not establish a compensable loss and that expected earnings must be proved where claimed.

Relevance

Multi-layer causation frequently produces speculative damages.

For example:

AI error → customer dissatisfaction → possible future customers lost

The claimant cannot automatically transform every possible future consequence into compensable damage.

17. Case 8 — Marwan Lutfi v DIFC Authority [2013] DIFC CA 003

The DIFC Court of Appeal discussed causation in the context of discriminatory treatment.

The Court explained that the causation inquiry concerns the reason why the relevant treatment occurred and that the causally operative factor must be identified.

Relevance

Although this is not a general UAE tort case, it demonstrates an important methodological point:

Causation requires identifying the factor that actually operated to produce the relevant consequence.

This approach can be adapted to complex automated decisions.

18. Direct Cause Versus Remote Cause

One of the most important distinctions is:

Direct cause

The defendant's act immediately produces the harm.

Example:

Driver negligently hits pedestrian → injury

Causal/consequential chain

The defendant's conduct operates through other events.

Example:

Defective software → wrong production schedule → factory shutdown → contractual penalty → financial loss

The second situation requires more sophisticated causal analysis.

Article 247 of the current Civil Transactions Law expressly recognises the distinction between direct harm and harm by causation.

19. Intervening Causes

An intervening cause is an event occurring between the defendant's conduct and the final injury.

Example:

Cybersecurity weakness

↓

Hacker intrusion

↓

AI system manipulation

↓

Fraudulent transaction

↓

Bank loss

Possible responsible parties could include:

• software provider;
• system operator;
• cybersecurity provider;
• hacker;
• employee.

The legal issue becomes whether the intervening event:

1. was foreseeable;
2. was independent;
3. substantially contributed to the damage;
4. completely displaced the original causal connection.

20. Concurrent Causes

Sometimes two independent causes combine.

Example:

Defective AI system + negligent employee

both contribute to the same financial loss.

Neither cause alone would have produced the final loss.

This is a concurrent causation problem.

The court may need to determine:

• whether both parties are responsible;
• whether responsibility can be apportioned;
• whether joint and several responsibility applies;
• whether one cause was merely incidental.

Article 253 of the current Civil Transactions Law is particularly relevant where several persons are responsible for the same harm.

21. Successive Causes

A different situation occurs where causes happen sequentially.

Example:

Developer defect

↓

Operator's negligent configuration

↓

Employee's failure to supervise

↓

Customer's improper use

↓

Damage

Here, several successive events may contribute.

The legal question becomes:

Did the later event merely contribute to the original harm, or did it replace the earlier event as the legally operative cause?

22. Multiple Agents and Causation

Multi-agent systems make this particularly complicated.

Suppose:

Agent A

collects data.

↓

Agent B

analyses data.

↓

Agent C

makes a recommendation.

↓

Agent D

executes transaction.

↓

Human

fails to supervise.

↓

Loss

There is no obvious single cause.

A court may need to ask:

23. Causation Mapping in Medical Negligence

Medical cases provide a useful practical analogy.

Layer 1

Incorrect diagnosis.

Layer 2

Incorrect treatment.

Layer 3

Failure to monitor.

Layer 4

Complication.

Layer 5

Permanent injury.

The claimant may need to prove not merely that a medical error occurred but that the error caused the injury.

The same logic can apply to:

• AI diagnostic tools;
• robotic surgery;
• automated hospital systems;
• electronic medical records.

24. Causation Mapping in Construction

Construction disputes commonly involve several layers:

Design defect

↓

Contractor deviation

↓

Subcontractor error

↓

Delay

↓

Additional cost

↓

Lost revenue

A claimant cannot simply attribute the entire loss to the first defect.

Each layer must be examined.

The BAM Higgs & Hill case demonstrates this principle particularly well because the Court rejected attempts to connect broad categories of claimed costs to alleged contractual breaches without establishing the necessary causal link.

25. Causation Mapping in Cybersecurity

A cyber dispute might look like:

Weak security

↓

Unauthorised access

↓

Data manipulation

↓

AI decision

↓

Fraudulent transaction

↓

Financial loss

Potential defendants could argue:

• the hacker was the sole cause;
• the attack was unforeseeable;
• the claimant failed to secure its own systems;
• the AI system merely responded to manipulated data.

The claimant would need to demonstrate the relevant causal chain.

26. Causation Mapping in Financial AI

Consider an automated trading system:

Incorrect market data

↓

AI prediction error

↓

Risk engine approves trade

↓

Trading agent executes

↓

Market moves

↓

Loss

The defendant might argue:

“The market movement was an independent cause.”

The claimant might argue:

“The trade would never have occurred without the defective AI output.”

This illustrates why but-for causation and substantial causation can become highly significant.

27. Causation and Foreseeability

A defendant may accept:

“My conduct contributed to the chain.”

but argue:

“The particular loss was too remote.”

This creates two distinct questions:

Factual causation

Did the conduct contribute to the loss?

Legal causation

Is the loss sufficiently connected to justify civil responsibility?

Multi-layer causation mapping should therefore never stop at factual causation.

28. Causation and Damages

A causal map should also be created for damages.

For example:

Wrongful act

↓

Immediate loss AED 100,000

↓

Business interruption AED 200,000

↓

Lost profit AED 500,000

↓

Reputational damage AED 1 million

↓

Future opportunity loss AED 2 million

Each category requires separate analysis.

The court may find:

• some losses directly caused;
• some consequential;
• some insufficiently proved;
• some too remote;
• some speculative.

29. Burden of Proof

Generally, the claimant must establish the factual foundation of the claim, including the damage and causal connection.

The claimant should therefore produce evidence capable of demonstrating:

1. the wrongful act;
2. the relevant event;
3. the causal mechanism;
4. the actual damage;
5. the relationship between the event and the damage.

In complex technological disputes, expert evidence can become especially important.

30. Expert Evidence

A multi-layer causation dispute may require experts in:

• software engineering;
• AI;
• cybersecurity;
• accounting;
• finance;
• medicine;
• construction;
• data science;
• telecommunications.

But the expert's role is primarily technical.

The court determines the legal consequence.

For example:

Expert: “Agent B's output increased the probability of the transaction by 85%.”

The court must determine:

“Does that establish legal causation under the applicable UAE law?”

Technical causation and legal causation are therefore distinct.

31. Causation and AI Explainability

AI creates a special problem because the system may not provide a simple explanation for its output.

A claimant may argue:

“The AI produced the wrong decision.”

But that alone may not prove:

“The defendant's legally actionable conduct caused my loss.”

The claimant may need to establish:

defect → AI output → reliance → action → damage

This is why logs, model versions, prompts, system architecture and audit records can become important evidence.

32. Breaking the Causal Chain

The causal chain may potentially be interrupted by:

1. Force majeure

An extraordinary external event.

2. Third-party conduct

An independent third party causes the damage.

3. Claimant's own conduct

The claimant materially contributes to the loss.

4. Independent market event

An unrelated market event produces the loss.

5. New contractual decision

The claimant voluntarily takes a new action that substantially changes the chain.

The court must assess the evidence rather than automatically accepting any intervening event as a complete defence.

33. Multi-Layer Causation and Contribution

A particularly useful distinction is:

Causal contribution

Did the person's conduct contribute to the harm?

Legal responsibility

Does the law impose liability for that contribution?

Apportionment

What share of responsibility should be assigned?

Recourse

Can one responsible person recover from another?

These are four different questions.

34. Practical Causation Matrix

A UAE court or lawyer could construct a table like this:

This is the essence of multi-layer causation mapping.

35. Causation Mapping and Article 253

Where several persons are legally responsible, the court can move from:

Causal mapping

to

Responsibility allocation

For example:

Developer — 30%

Operator — 40%

Supervisor — 20%

Claimant — 10%

This is only an illustrative analytical model, not a statement that UAE courts must use those percentages in a particular case.

The actual allocation depends on the applicable law and evidence.

36. Causation in Contract and Tort

Multi-layer causation can arise under both:

Contractual liability

Example:

AI vendor fails to perform contractual specifications.

Tort liability

Example:

Defective automated system causes damage to a non-contracting third party.

Concurrent contractual and tortious conduct

Example:

A bank's AI vendor breaches its contract with the bank and the defective system also causes damage to a customer.

The applicable legal regime must therefore be identified before constructing the causal map.

37. Important Difference Between Cause and Condition

Not every condition leading to damage is necessarily a legal cause.

Example:

A company installs AI software.

The software operates correctly.

An unrelated power outage shuts down the business.

The installation of AI may be a condition in the background, but not necessarily the legal cause of the loss.

Therefore:

Chronological sequence does not automatically equal legal causation.

38. Important Difference Between Cause and Opportunity

Suppose an AI system contains a vulnerability.

A hacker later exploits it.

The vulnerability may have created the opportunity for the attack.

But the court must still determine:

• whether the vulnerability was legally significant;
• whether the attack was foreseeable;
• whether the hacker's conduct was an independent cause;
• whether the defendant had a duty to prevent the vulnerability.

This is a classic multi-layer causation problem.

39. Multi-Layer Causation in Digital Platforms

Consider:

Platform algorithm

↓

Content recommendation

↓

User exposure

↓

User conduct

↓

Third-party transaction

↓

Financial loss

The platform may argue that the user's voluntary conduct was the intervening cause.

The claimant may argue that the algorithm materially caused the conduct.

The legal question becomes highly fact-specific.

The court must distinguish:

• mere background conditions;
• substantial causal factors;
• intervening causes;
• legally attributable consequences.

40. Key Principles for UAE Law

Principle 1

Causation is an essential component of civil liability.

Principle 2

UAE law recognises direct and causal/consequential harm.

Principle 3

Multiple actors can contribute to one harmful result.

Principle 4

A breach alone does not establish entitlement to damages.

Principle 5

The claimant must connect the claimed loss to the wrongful conduct.

Principle 6

Intervening acts can affect the causal chain.

Principle 7

Lost profits require adequate proof and causal connection.

Principle 8

Modern technological disputes may contain several causal layers rather than one simple cause.

Principle 9

Technical causation and legal causation are separate inquiries.

Principle 10

Multi-layer causation mapping should precede allocation of liability and assessment of damages.

41. Six+ Case Laws — Quick Revision Table

42. Conclusion

Multi-layer causation mapping in UAE civil law is a structured method for identifying how multiple acts, omissions, technologies, human decisions and external events combine to produce legally compensable damage.

The current Civil Transactions Law is particularly significant because Article 247 expressly recognises the distinction between direct harm and harm by causation, while the provisions dealing with multiple responsible persons provide the framework for addressing situations where responsibility is distributed across several actors.

The central analytical formula is:

Conduct → Intermediate Event → Intervening Event → Final Harm → Legal Causation → Responsibility → Compensation

For AI and digital disputes, this can be expanded to:

Data → Algorithm → Agent → Human Decision → Automated Execution → External Event → Damage

The important legal lesson is that the first event in a chain is not automatically the legal cause, and the final event is not automatically the sole cause. UAE civil-law analysis requires the court to identify the legally relevant causal connection and then determine responsibility and compensation on the basis of the proven damage.

Note: The DIFC judgments cited above are UAE-based authorities, but DIFC decisions are not automatically binding precedents on UAE mainland courts. Where a case discusses the former UAE Civil Code, it should also be read in light of the current Federal Decree-Law No. 25 of 2025, effective from 1 June 2026.