Civil Law And Distributed Legal Networks Blockchain Based Civil Law Structures In Uae .
1. Introduction
Distributed liability in autonomous economic agent networks is an emerging area of civil law concerned with determining legal responsibility when multiple artificial intelligence (AI) systems, autonomous software agents, smart contracts, blockchain platforms, and automated commercial entities interact to perform economic activities without continuous human supervision.
In the United Arab Emirates (UAE), this issue is particularly important because of the country's development of:
Artificial intelligence and autonomous digital services.
Blockchain-based commercial transactions.
Smart contracts and automated payment systems.
Digital banking and fintech platforms.
Algorithmic trading and automated investment systems.
Decentralised autonomous organisations (DAOs).
Machine-to-machine commercial transactions.
The central legal question is: When an autonomous economic agent causes financial loss, breaches a contract, makes an unauthorised payment, or damages another party through its interaction with other autonomous agents, who should bear civil liability?
Possible responsible parties include the developer, software owner, deploying business, operator, data provider, platform provider, cybersecurity service provider, and the human or corporate principal that authorised the agent.
The UAE's existing civil-law framework provides important foundations for addressing these disputes, even though it does not establish a comprehensive, standalone regime for distributed liability among autonomous economic agents.
2. Meaning and Nature of Autonomous Economic Agent Networks
An autonomous economic agent is a software-based system capable of independently performing commercial or financial activities according to programmed objectives, permissions, data inputs, and decision-making rules.
Examples include:
AI agents negotiating commercial contracts.
Autonomous systems purchasing goods and services.
Algorithmic trading agents executing financial transactions.
Smart-contract agents distributing payments.
AI procurement agents selecting suppliers.
Autonomous insurance-claim processing systems.
Decentralised agents managing digital assets.
Machine-to-machine payment networks.
2026 Multiagent Architecture Stack - $79 Website Template | Lucky Graphics | Lucky Graphics
Sociedades Automatizadas y DAOs - Facultad de Ciencias JurΓdicas y Sociales
10
Important characteristics
A. Distributed decision-making
Different agents may perform separate functions in one commercial transaction.
B. Independent execution
An agent may initiate payments, purchases, or contractual actions without obtaining fresh human approval for every transaction.
C. Interdependence
The conduct of one agent may influence the decisions and actions of other agents.
D. Multiple control points
Different organisations may control the software, infrastructure, datasets, algorithms, and execution mechanisms.
E. Difficulty in identifying fault
A financial loss may result from several interacting technical failures rather than one identifiable human decision.
This creates a significant challenge for traditional civil liability principles.
3. Legal Framework Governing Distributed Liability in the UAE
The UAE legal framework can be understood through several interconnected laws.
A. UAE Civil Transactions Law β Federal Law No. 5 of 1985
The Civil Transactions Law is the principal foundation for civil compensation and liability.
Its important provisions include:
| Article | Legal principle | Application to autonomous agents |
|---|---|---|
| 282 | General liability for harm | Liability for damage caused through an autonomous system |
| 283 | Direct and consequential harm | Identifying direct and indirect technical contributors |
| 284 | Priority of direct responsibility | Determining responsibility where direct and indirect causes overlap |
| 285 | Liability for deception | Fraudulent AI-generated instructions or representations |
| 287 | External causes and force majeure | Cyberattacks, third-party interference and unavoidable failures |
| 289 | Responsibility of the actor | Examining the relationship between the person executing an act and the person ordering it |
| 291 | Multiple wrongdoers | Allocation of liability among several responsible participants |
| 292 | Compensation for actual damage and lost profits | Calculating financial losses caused by agent networks |
| 293 | Moral damage | Compensation where legally recognised non-financial harm occurs |
Article 291 is particularly important. It provides that when several persons are responsible for a harmful act, each is liable according to their contribution, and the judge may impose equal, joint, or several liability.
This provides a legal foundation for addressing situations in which multiple businesses contribute to a single loss through interconnected autonomous systems.
DIFC Courts+1
B. Electronic Transactions and Trust Services Law β Federal Decree-Law No. 46 of 2021
This legislation is especially relevant to autonomous commercial transactions.
Article 11 recognises contracts made between automated electronic systems, including contracts concluded without direct human intervention.
This is important because an autonomous economic agent may:
Make an electronic offer.
Accept another automated system's offer.
Execute a purchase.
Confirm a payment.
Complete an electronic commercial transaction.
Article 12 addresses attribution of electronic documents to their originators.
Article 18 establishes the admissibility and legal effect of electronic evidence, subject to the statutory requirements.
Consequently, a business cannot automatically avoid contractual responsibility simply by arguing that a machine rather than a human employee executed the transaction.
u.ae+1
C. Commercial and contractual principles
Contractual liability may arise when an organisation fails to perform obligations associated with its autonomous agent network.
Examples include:
Failure to execute authorised transactions.
Executing payments beyond agreed limits.
Incorrect supplier selection.
Failure to maintain cybersecurity protections.
Breach of automated procurement agreements.
Failure to supervise outsourced AI systems.
The contractual relationship between the parties, the applicable legislation, and the agreed allocation of risk will determine the available remedies.
D. Data protection and cybersecurity obligations
Autonomous economic agents depend on data and connected infrastructure.
A compromised dataset, stolen authentication credential, or manipulated API can cause an agent to perform an unauthorised transaction.
Relevant legislation includes the UAE Personal Data Protection Law, Federal Decree-Law No. 45 of 2021, and applicable cybersecurity, financial-sector and sector-specific regulations.
A data protection violation may create regulatory consequences and, where the requirements of civil liability are established, a separate claim for compensation.
4. Major Categories of Distributed Civil Liability
A. Developer liability
Developers design the algorithms, decision-making structures, and operational rules of autonomous agents.
They may face liability where their negligence or contractual breach causes foreseeable harm.
Examples:
Defective transaction algorithms.
Inadequate testing before commercial deployment.
Failure to correct known vulnerabilities.
Unsafe autonomous payment mechanisms.
Inadequate safeguards against manipulation.
Failure to implement appropriate transaction limits.
Example: An AI developer creates a procurement agent that repeatedly purchases goods at artificially inflated prices because its bidding algorithm contains a defective pricing instruction. If the developer's actionable fault and causal contribution are established, civil liability may arise.
B. Operator liability
The operator is the person or organisation deploying and controlling the agent network in commercial activities.
Operator responsibility may arise from:
Negligent configuration.
Inadequate human supervision.
Failure to establish transaction limits.
Ignoring system warnings.
Continuing to use a known defective system.
Failure to suspend an agent after detecting suspicious activity.
The operator may be particularly important because it determines the commercial purpose for which the network is deployed.
C. Platform-provider liability
A platform provider supplies the infrastructure through which autonomous agents communicate and execute transactions.
Potential failures include:
Incorrect routing of transactions.
Failure of authentication mechanisms.
Inadequate access controls.
Incorrect execution of smart contracts.
Failure to preserve transaction records.
Platform outages causing commercial losses.
Liability depends on the provider's contractual obligations, applicable duties, fault, and causal contribution.
D. Data-provider liability
Autonomous economic agents depend on reliable information.
A data provider may contribute to damage by supplying inaccurate, manipulated, incomplete, or unlawfully obtained information.
For example, an autonomous trading agent receives incorrect market information from an external data provider and purchases securities at an unreasonable price.
The data provider may be liable if the relevant legal requirements are satisfied and the loss is attributable to its wrongful conduct.
E. Principal and corporate liability
A company that authorises autonomous agents to conduct business may be responsible for transactions performed within the scope of their authorised functions.
Relevant questions include:
Who owns the agent?
Who authorised its commercial activities?
What transaction limits were imposed?
Who controls the agent's credentials?
Who receives the economic benefits?
Who is responsible for supervision?
A company should not be able to avoid its contractual obligations merely because its authorised commercial activity was performed automatically.
However, this does not mean that every developer or infrastructure provider automatically becomes liable for every transaction.
F. Cybersecurity and malicious-interference liability
A third party may hack, manipulate, or take control of an autonomous agent network.
Article 287 may become relevant where the defendant establishes that the damage resulted from an external cause in which it played no part, subject to applicable legal provisions and agreements.
However, an organisation cannot necessarily rely on an external cyberattack if its own security failures materially contributed to the loss.
5. Distributed Liability and the Role of Article 291
Article 291 is the central provision for analysing multiple responsible participants.
Consider the following commercial network:
Suppose five participants are involved:
| Participant | Function | Possible liability |
|---|---|---|
| Developer | Creates the AI agent | Defective software or negligent design |
| Operator | Deploys the agent | Negligent supervision or configuration |
| Data provider | Supplies information | Incorrect or manipulated data |
| Platform provider | Executes transactions | Infrastructure or execution failures |
| Commercial principal | Authorises transactions | Contractual or operational responsibility |
Suppose the network executes an unauthorised AED 5 million transaction.
The court would need to examine:
Whether the transaction was authorised.
Whether the software was defective.
Whether the operator failed to supervise the system.
Whether the data provider supplied incorrect information.
Whether the platform executed the instruction correctly.
Whether an external attacker intervened.
Whether each participant's conduct caused or contributed to the loss.
Article 291 allows the court to consider the respective contributions and impose joint or several liability where appropriate.
Important distinction: Article 291 does not automatically make every participant jointly liable. The court must establish the relevant responsibility and determine the appropriate allocation under the facts.
6. Contractual Liability Between Autonomous Agents
A particularly difficult issue arises when two autonomous agents conclude a contract without direct human intervention.
Under Article 11 of Federal Decree-Law No. 46 of 2021, automated electronic contracting is legally recognised.
For example:
Agent A represents a manufacturing company.
Agent B represents a supplier.
Agent A automatically requests 10,000 units.
Agent B automatically accepts the offer.
A smart contract initiates payment.
A delivery system arranges shipment.
If Agent A mistakenly orders 100,000 units instead of 10,000, the following questions arise:
Was the automated order within the authority granted by the principal?
Did the supplier know or have reason to know that the system was automated?
Was there a software malfunction?
Did the supplier act in good faith?
Does the contract contain an error-correction mechanism?
Who bears the contractual risk of the automated mistake?
The legal analysis should distinguish between a valid automated contract, a defect in contractual consent, an unauthorised transaction, and a failure to perform an existing contractual obligation.
7. Sixteen Important Case Laws and Judicial Authorities
Preliminary qualification: There is no established body of UAE reported judgments specifically deciding the allocation of civil liability between multiple autonomous AI economic agents. The following cases are relevant because they establish the underlying principles of causation, multiple wrongdoers, corporate responsibility, fraud, and compensation. Their application to autonomous networks is analytical and analogous, not a claim that the courts have already decided AI-agent liability.
Case 1. Abu Dhabi Court of Cassation β Civil Appeals Nos. 208 and 214 of 2023
Judgment: 6 December 2023
Legal principle: Multiple wrongdoers and joint liability.
The court addressed the legal consequences of multiple persons being responsible for harmful conduct.
The judgment is important for understanding Article 291 and the circumstances in which several wrongdoers may be held jointly liable.
Application to autonomous agents: Where an operator, software developer, and platform provider contribute to one harmful transaction, the court may examine their respective contributions and the possibility of joint or several liability.
Polaris Legal Library+1
Case 2. Abu Dhabi Court of Cassation β Civil Appeal No. 96 of 2024
Legal principle: Fault, damage, causation and the relationship between criminal and civil responsibility.
The case concerned serious workplace injuries and the relationship between employer fault, established criminal responsibility, and civil compensation.
Application: A company deploying autonomous machinery or AI-controlled economic infrastructure may face civil liability when its own operational failures cause foreseeable harm. Criminal and regulatory findings may also have evidentiary significance, subject to the applicable procedural rules.
Polaris Legal Library
Case 3. Abu Dhabi Court of Cassation β Civil Appeal No. 1196 of 2023
Judgment: 24 January 2024
Legal principle: Harmful conduct includes exceeding the required limits of conduct or failing to meet the required standard, whether through action or omission.
Application: An organisation may face liability for failing to install appropriate monitoring mechanisms, transaction controls, or cybersecurity safeguards where that omission causes damage.
AssetType Images
Case 4. Abu Dhabi Court of Cassation β Commercial Appeal No. 1104 of 2021
Legal principle: Contractual liability requires fault, damage and a causal relationship.
The judgment discusses compensation for loss and lost profits resulting from harmful contractual conduct.
Application: If an autonomous procurement system fails to perform a valid commercial agreement, the claimant must establish the relevant contractual breach, actual damage, and causal connection.
adgm.com
Case 5. Abu Dhabi Court of Cassation β Commercial Appeal No. 700 of 2009
Legal principle: Compensation for actual losses and loss of earnings.
The judgment discusses the operation of Articles 282, 292 and 293 of the Civil Transactions Law.
Application: An autonomous economic network that causes commercial disruption may expose responsible parties to claims for actual financial losses and legally recoverable lost profits.
adgm.com
Case 6. Dubai Court of Cassation β Commercial Appeals Nos. 46 and 49 of 2006
Legal principle: Compensation must relate to established damage, including reasonably substantiated lost earnings.
The court's approach distinguishes speculative expectations from financial losses supported by reasonable grounds.
Application: A business claiming losses from an autonomous trading or payment network must provide reliable evidence of its actual losses and sufficiently established lost profits.
DIFC Courts
Case 7. Dubai Court of Appeal β Civil Appeal No. 1477 of 1999
Legal principle: Actual loss and reasonably expected lost profits may be compensable.
Application: A claimant affected by the failure of an autonomous commercial system may seek compensation for profits that would reasonably have been earned, provided the necessary evidence and causation are established.
DIFC Courts
Case 8. Dubai Court of Cassation β Commercial Appeal No. 151 of 2025
Legal principle: Assessment of compensation under Article 292.
The case is cited in connection with the assessment of actual harm and loss of profit.
Application: It is relevant to calculating damages where automated commercial activities cause financial injury.
Polaris Legal Library
Case 9. Abu Dhabi Court of Cassation β Commercial Appeal No. 1137 of 2023
Legal principle: Compensation under Article 292 for damage and consequential lost profits.
Application: This principle is relevant when an autonomous agent's wrongful execution of transactions produces demonstrable financial losses.
Polaris Legal Library
Case 10. Dubai Court of Cassation β General Assembly Case No. 5 of 2022
Legal principle: Assessment of compensation under the Civil Transactions Law.
Application: It provides relevant background for determining the scope and calculation of damages in complex commercial liability disputes.
Polaris Legal Library
Case 11. Larmag Holding B.V. v First Abu Dhabi Bank PJSC and Others [2019] DIFC CFI 054
Legal principle: UAE Civil Code Article 282 establishes a general civil wrong, while Articles 283, 285 and 292 address causation, deception and compensation.
The judgment is significant because it considers the structure of UAE civil-law remedies and the importance of properly identifying and pleading the relevant legal causes of action.
Application: In an autonomous-agent dispute, a claimant should identify the particular wrongdoing of each participant rather than simply alleging that the entire network malfunctioned.
DIFC Courts+1
Case 12. Abu Dhabi Commercial Bank PJSC v Shetty and Others [2022] EWHC 529 (Comm)
Legal principle: UAE Civil Code provisions concerning harmful conduct, deception, multiple wrongdoers and compensation.
The English Commercial Court considered claims framed under UAE civil law, including Articles 282, 285, 291 and 292.
Application: The case provides useful comparative judicial analysis of multiple participants allegedly involved in coordinated harmful conduct. It is relevant when evaluating coordinated actions among autonomous agents and their human or corporate controllers.
Bailii+1
Case 13. Globemed Gulf Healthcare Solutions LLC v Oman Insurance Company PSC [2017] DIFC CFI 051
Legal principle: Compensation requires established damage; speculative or merely possible injury is insufficient.
The judgment discusses UAE Court of Cassation authorities on the assessment of actual damage and lost earnings.
Application: A business alleging losses caused by an autonomous economic network must establish that the loss occurred or will certainly occur, rather than relying on hypothetical future losses.
DIFC Courts
Case 14. Union Properties PJSC and Another v Trinkler & Partners Ltd and Others [2026] ADGMCFI 0010
Judgment: 6 March 2026
Legal principle: Multiple persons responsible for harmful conduct may be liable under Article 291 in proportion to their respective contributions, with the possibility of joint or several liability.
The judgment considers the interaction between UAE Civil Code principles and claims involving coordinated conduct.
Application: It is particularly relevant by analogy to autonomous networks in which multiple independent commercial participants cooperate in a transaction that causes loss.
Bailii
Case 15. Emirates NBD Bank PJSC v Almakhawi and Others [2025] DIFC CFI 039
Legal principle: UAE Civil Code provisions governing harmful acts, direct and consequential harm, deception, multiple wrongdoers and compensation.
The judgment considers the legal consequences of coordinated asset transfers and alleged conduct intended to prejudice creditors.
Application: It illustrates how a court can examine a network of participants and their coordinated transactions rather than treating each transaction as an isolated event. This is useful when analysing distributed autonomous financial transactions.
DIFC Courts+1
Case 16. Union Properties PJSC v Trinkler and related authorities on UAE-law causation
Legal principle: The distinction between conduct, causal contribution and compensable damage must be established under the applicable legal framework.
Application: In an AI-agent dispute, technical involvement alone should not automatically establish civil liability. The claimant must demonstrate the relevant legal basis for attributing harm to the defendant.
This is a supplementary application of the principles considered in Case 14, not a separate judgment.
8. Allocation of Liability Between Network Participants
A major issue in autonomous economic networks is how responsibility should be distributed when several participants contribute to the same loss.
A useful analytical model is the following:
| Responsible participant | Relevant legal question | Potential consequence |
|---|---|---|
| AI developer | Was the algorithm negligently designed? | Developer liability |
| System owner | Were adequate safeguards provided? | Owner liability |
| Commercial operator | Was the system negligently deployed? | Operator liability |
| Data supplier | Was inaccurate information supplied? | Data-related liability |
| Cloud provider | Did infrastructure failure cause damage? | Contractual or civil liability |
| Smart-contract provider | Was execution defective? | Contractual or civil liability |
| Human supervisor | Were warnings ignored? | Personal liability where established |
| Corporate principal | Was the transaction authorised? | Contractual responsibility |
Proportionate liability
Under Article 291, the court may consider the contribution of each responsible participant.
For example, assume a financial loss of AED 10 million arises from an interconnected autonomous trading network.
The technical investigation establishes that:
The developer introduced a defective algorithm.
The operator failed to apply transaction restrictions.
The data provider supplied manipulated information.
A third-party attacker exploited an authentication weakness.
The court would investigate each party's conduct and the causal relationship between that conduct and the loss.
The percentages of responsibility cannot simply be presumed from the technical findings. They must be established through evidence and the applicable legal principles.
Joint and several liability
Article 291 also permits the judge to order joint or several liability.
This is particularly relevant where multiple wrongdoers contribute to one indivisible harmful result.
The injured party may potentially obtain a judgment against more than one responsible participant, subject to the terms of the judgment and applicable contribution or recourse rules.
9. Liability for Autonomous Financial Transactions
Autonomous economic agents may be used in banking, fintech, digital investment, and payment services.
#ai #finance #quant | Yves Hilpisch
Cosmic Nexcus | Autonomous Banking Powered by Agentic AI
8
Consider the following examples.
Example 1: Algorithmic trading failure
An autonomous trading agent executes thousands of transactions because of a defective risk-management instruction.
Possible liability:
The developer for an actionable software defect.
The operator for negligent deployment or supervision.
Other providers if their wrongful conduct contributed to the loss.
Example 2: Unauthorised payment execution
A payment agent transfers funds to an unauthorised recipient after its authentication credentials are compromised.
Relevant issues include:
Whether the payment was authorised.
Whether the system followed the agreed authentication process.
Whether the platform complied with its contractual and regulatory obligations.
Whether the account holder contributed to the loss.
Whether a third party committed fraud.
Example 3: Autonomous insurance processing
An insurance agent incorrectly rejects legitimate claims because of defective automated decision-making.
The insurer may face contractual liability where the rejection constitutes a breach of its obligations. The developer's responsibility would require a separate examination of its duties and conduct.
10. Smart Contracts and Distributed Liability
Smart contracts create additional difficulties because contractual performance may occur automatically.
For example, a decentralised commercial network may automatically:
Receive a purchase instruction.
Verify payment.
Release digital assets.
Calculate contractual penalties.
Transfer funds to suppliers.
Update the transaction ledger.
If the smart contract executes an incorrect instruction, the following questions arise:
Was the code consistent with the commercial agreement?
Was the programming defective?
Did a participant supply incorrect information?
Was the execution authorised?
Could the transaction have been suspended?
Was there a failure to implement reasonable security controls?
An important distinction must be maintained between the legal validity of an electronic transaction and the civil liability arising from its defective execution.
Article 11 of the Electronic Transactions and Trust Services Law supports automated contracting. It does not, however, give autonomous software an independent legal personality or automatically immunise the people and companies behind it.
11. Evidence and Technical Attribution of Liability
Distributed autonomous networks create significant evidentiary difficulties.
A claimant must often establish what happened inside several interacting systems.
The following evidence may be particularly important:
| Evidence | Legal significance |
|---|---|
| Source code | Identifies programming defects |
| System logs | Establishes operational events |
| Blockchain records | Shows recorded transactions |
| API records | Establishes communication between agents |
| Audit trails | Identifies system changes |
| Access-control records | Helps identify authorised users |
| Model versions | Identifies algorithmic changes |
| Training-data records | Helps identify defective information |
| Cybersecurity reports | Establishes possible intrusion |
| Expert reports | Explains technical causation |
A. Electronic evidence
Federal Decree-Law No. 46 of 2021 recognises electronic documents and electronic transactions as legally admissible evidence, subject to the applicable statutory conditions.
B. Expert evidence
An expert may be required to determine:
Which agent initiated the transaction.
Which instruction triggered the harmful result.
Whether the system behaved according to its specifications.
Whether an external attacker intervened.
Whether a software defect existed.
Whether the loss could have been prevented.
C. Blockchain evidence
Blockchain records can help establish the timing and sequence of transactions.
However, a blockchain record does not necessarily prove:
The real-world identity of the person controlling an account.
The truth of the information originally entered.
The legal authority of the person issuing the instruction.
The absence of manipulation before recording.
Therefore, blockchain evidence should be assessed together with authentication records, contractual documents and expert evidence.
12. Defences Available to Network Participants
Participants may rely on several legal defences, depending on the facts.
A. Absence of fault or legal responsibility
A developer may argue that the software performed according to the agreed specifications and that the loss arose from an independent external cause.
B. Force majeure and external causes
Article 287 recognises certain external causes, including force majeure, unavoidable accidents, third-party acts and the conduct of the injured person.
C. Victim contribution
Where the claimant's own conduct caused or contributed to the damage, the court may consider its legal significance.
D. Lack of causation
A participant may establish that its conduct did not cause the alleged damage.
E. Contractual allocation of risk
Contracts may allocate responsibility for system failures, data accuracy, security breaches and transaction errors, subject to mandatory legislation and applicable legal limitations.
F. Unauthorised third-party interference
A defendant may argue that a hacker or unauthorised user caused the loss. However, the defence may fail to exclude responsibility if the defendant's own negligence materially contributed to the incident.
13. Remedies Available to Injured Parties
The injured party may seek different remedies depending on whether the claim is contractual, non-contractual, or based on a combination of legal grounds.
| Remedy | Purpose |
|---|---|
| Compensatory damages | Recover actual financial losses |
| Lost-profit compensation | Recover sufficiently established lost earnings |
| Contractual damages | Address failure to perform contractual obligations |
| Restitution or repayment | Recover money where an applicable legal basis exists |
| Injunctions | Prevent continuing harmful conduct where available |
| Declaration of rights | Clarify contractual or legal responsibility |
| Interim protective measures | Preserve assets and evidence where legally available |
Article 292 is particularly relevant to compensation for actual damage and lost profits that are a natural result of the harmful act.
Article 293 may also be relevant where legally recognised moral damage is established.
14. Jurisdiction and Cross-Border Liability
Autonomous economic agent networks often operate across different countries.
For example:
The developer may be based in Europe.
The cloud infrastructure may be located in the United States.
The commercial operator may be incorporated in Dubai.
The data provider may be based in Singapore.
The injured customer may be located in Abu Dhabi.
This creates questions concerning jurisdiction, applicable law, contractual choice-of-law provisions, and enforcement.
UAE jurisdictional considerations
Depending on the parties and the nature of the transaction, proceedings may involve:
UAE onshore civil courts.
Dubai International Financial Centre (DIFC) Courts.
Abu Dhabi Global Market (ADGM) Courts.
Arbitration where a valid arbitration agreement exists.
The DIFC and ADGM have distinct legal frameworks and should not be treated as identical to the UAE onshore civil-law system.
A court must determine its jurisdiction and the applicable substantive law before deciding the liability dispute.
15. Regulatory Challenges and Future Legal Development in the UAE
The existing legal framework provides important principles, but autonomous economic agent networks introduce issues that traditional civil liability rules were not specifically designed to address.
Several developments would improve legal certainty.
A. Mandatory identification of autonomous agents
Commercially deployed agents could be required to maintain:
Unique digital identifiers.
Identifiable corporate principals.
Defined transaction authority.
Traceable execution records.
Clear ownership and operational information.
B. Mandatory human accountability
Businesses should maintain identifiable human or corporate responsibility for high-risk autonomous transactions.
Delegating commercial decisions to software should not eliminate accountability.
C. AI liability insurance
The UAE could encourage insurance products covering:
Algorithmic trading losses.
Autonomous procurement errors.
Smart-contract failures.
AI-related commercial negligence.
Cybersecurity incidents.
Multi-agent network failures.
D. Mandatory technical audit systems
High-risk autonomous economic networks could be subject to independent technical audits covering:
Algorithmic accuracy.
Cybersecurity.
Data integrity.
Transaction limits.
Human oversight.
Incident reporting.
E. Distributed responsibility framework
A future specialised framework could clarify how courts distinguish between:
Design responsibility.
Deployment responsibility.
Operational responsibility.
Data responsibility.
Infrastructure responsibility.
Supervisory responsibility.
F. Autonomous agent liability registry
A centralised or regulated digital registry could identify the responsible commercial principal, authorised operators and relevant technical service providers.
Such a system would help injured parties identify the appropriate defendants.
16. Practical Case Study: Autonomous Commercial Network Failure in Dubai
Consider a hypothetical Dubai-based company using five autonomous agents to manage its international supply chain.
Agentic AI: Redefining the Future of Supply Chain and Logistics
The agents perform the following functions:
Agent A identifies suppliers.
Agent B negotiates prices.
Agent C approves purchase orders.
Agent D executes payments.
Agent E manages delivery.
A programming error causes Agent B to negotiate an incorrect price. Agent C approves the order, and Agent D transfers AED 8 million to an unintended recipient.
The company suffers financial losses.
Legal analysis
Step 1: Identify the harmful act
The court examines whether the transaction resulted from a defective algorithm, negligent operation, unauthorised instruction, or another wrongful act.
Step 2: Identify the responsible participants
The developer, operator, payment provider and any other relevant participants are examined.
Step 3: Establish causation
Technical experts reconstruct the communication and transaction sequence.
Step 4: Apply the Civil Transactions Law
Articles 282 and 283 provide the general liability framework. Article 291 becomes relevant where multiple persons are responsible for the harmful act.
Step 5: Determine contractual responsibility
Article 11 of Federal Decree-Law No. 46 of 2021 is relevant to the validity of the automated electronic contract.
Step 6: Calculate compensation
Article 292 provides the basis for assessing established damage and qualifying lost profits.
Step 7: Determine allocation
The court determines the responsibility of each participant and whether proportionate, joint or several liability is appropriate.
17. Important Legal Principles
The following principles summarise the legal position:
| Principle | Significance |
|---|---|
| Technological neutrality | Civil liability can apply even when harmful conduct is performed through software |
| Human and corporate accountability | Autonomous execution does not automatically eliminate the principal's responsibility |
| Causation | Technical participation alone is insufficient to establish liability |
| Multiple wrongdoers | Article 291 accommodates multiple responsible participants |
| Contractual attribution | Automated contracts can be legally effective |
| Evidence integrity | Electronic records must be properly authenticated and assessed |
| Proportionate responsibility | Individual contributions may influence liability allocation |
| Compensation | Actual damage and sufficiently established lost profits may be recoverable |
18. Conclusion
Distributed liability in autonomous economic agent networks represents an important emerging development in UAE civil law.
The UAE Civil Transactions Law, particularly Articles 282, 283, 284, 287, 289, 291 and 292, provides the principal legal foundation for determining responsibility when autonomous systems cause commercial or financial damage.
Federal Decree-Law No. 46 of 2021 adds an important dimension by recognising contracts concluded through automated electronic systems, even without direct human intervention.
The most significant provision for distributed liability is Article 291, because it allows courts to examine the responsibility of multiple wrongdoers and determine whether liability should be proportionate, joint or several.
However, the existence of an interconnected AI network does not itself establish joint liability. Courts must examine the conduct, legal duties, causation and responsibility of each relevant participant.
The central conclusion is that UAE civil law can accommodate many disputes involving autonomous economic agent networks through existing principles of contractual responsibility, harmful acts, causation and compensation, although a dedicated statutory framework for multi-agent AI liability would improve legal certainty and commercial protection.
Research note: The 15 distinct judicial authorities discussed above are relevant to the existing UAE civil-law principles. They should not be represented as 15 decisions specifically concerning autonomous AI agents. Some are DIFC or ADGM decisions, and some are foreign judgments applying or examining UAE law. The precise applicability of each authority depends on the forum and governing law. This distinction is important for academic research and legal submissions.

comments