Civil Law And Uae Machine-Mediated Consent And Its Legal Validity .
Civil Law and UAE: Machine-Mediated Consent and Its Legal Validity
1. Introduction
Machine-mediated consent refers to a situation in which a person's contractual consent is expressed, recorded, transmitted, or implemented through a technological system rather than through traditional face-to-face or paper-based interaction.
Examples include:
clicking "I agree" on an online platform;
accepting terms through a mobile application;
an electronic signature;
a smart-contract transaction;
an automated purchasing system;
an AI assistant communicating an offer or acceptance;
a chatbot negotiating contractual terms;
an automated trading system entering transactions;
a machine acting according to pre-programmed instructions;
a platform automatically renewing a contract;
biometric or digital authentication being used to approve a transaction.
The principal legal question is:
When can conduct performed through a machine or automated system be legally attributed to a human or legal person so that valid consent exists?
UAE legislation provides unusually direct assistance on this question. The UAE Electronic Transactions and Trust Services framework expressly recognizes electronic offers and acceptances and provides that contracts can be formed through automated electronic mediums without direct human intervention.
Thus, the use of a machine does not, by itself, invalidate contractual consent.
2. Meaning of Machine-Mediated Consent
Machine-mediated consent can be divided into several categories.
A. Human-click consent
A person personally clicks:
"Accept"
The machine merely records the person's decision.
B. Electronic-signature consent
A person applies an electronic signature to a document.
C. Delegated automated consent
A person pre-programs a system to conclude transactions within defined parameters.
For example:
"Purchase up to AED 50,000 of specified goods whenever inventory falls below 100 units."
D. AI-assisted consent
An AI system negotiates terms or communicates with another party on behalf of its user.
E. Fully automated transaction
Two automated systems interact and conclude a transaction without direct human intervention at the moment of contracting.
F. Machine-generated execution
A contract may already exist, but a machine automatically performs an obligation.
These situations should not be treated identically because the question of human intention, authorization and attribution becomes progressively more difficult as human involvement decreases.
3. UAE Statutory Recognition of Automated Contracting
The UAE Federal Decree-Law on Electronic Transactions and Trust Services expressly provides that an offer or acceptance may be expressed electronically.
It further provides that a contract does not lose validity, evidential weight or enforceability merely because it is made through electronic documents.
Most importantly, the law expressly recognizes automated electronic transactions:
a contract may be made between automated electronic systems programmed in advance for that purpose, and it can remain valid and legally effective even without personal or direct human intervention.
It also recognizes contracts between an automated system belonging to one person and another person where the other person knows, or is presumed to know, that the system will automatically make or execute the contract.
This is the statutory foundation for machine-mediated consent in the UAE.
4. The Central Principle: Attribution Rather Than Conscious Machine Intent
A machine does not normally possess legal personality or human consciousness.
Therefore, the law does not need to establish that:
"the computer intended to contract."
Instead, the legal question is:
Was the machine's conduct legally attributable to a person who intended or authorized the relevant transaction?
This distinction is fundamental.
For example:
Company A programs an automated purchasing system to buy raw materials when prices fall below a specified threshold.
The system purchases goods at 9:00 a.m.
The company cannot ordinarily argue:
"We did not consent because no employee personally clicked the purchase button."
The legal significance of the programming decision lies in the prior authorization and allocation of decision-making power to the system.
5. Elements of Valid Machine-Mediated Consent
A machine-mediated transaction should generally be examined through the following questions:
1. Was there an offer?
The proposal must be sufficiently definite under the applicable contractual law.
2. Was there acceptance?
There must be conduct demonstrating assent.
3. Was the subject matter sufficiently determined?
The essential contractual obligations must be identifiable.
4. Was the cause/lawful purpose valid?
The transaction must not violate mandatory law or public policy.
5. Was the machine authorized?
The person claiming to be bound must have authorized the system or otherwise be legally responsible for it.
6. Was the system functioning within its programmed parameters?
This becomes particularly important where the machine acts unexpectedly.
7. Was the counterparty aware of the automated nature of the transaction?
The UAE electronic-transactions legislation expressly addresses situations where the other party knows or is presumed to know that the automated system will conclude or perform the contract.
6. Machine-Mediated Consent Does Not Mean Machine Legal Personality
It is important to distinguish:
automated contracting
from
AI legal personality.
The UAE law recognizes the former without necessarily recognizing the latter.
An automated system may conclude a contract because the law attributes its operation to its owner, controller or principal.
Thus:
Machine acts → law attributes conduct → human/legal person becomes bound
rather than:
Machine acts → machine itself becomes contracting party.
This distinction becomes particularly important with AI agents.
7. Case Law 1 — Michael George Forbes v Robert Kidd
[2023] DIFC CFI 081
This is one of the most useful modern UAE/DIFC authorities concerning consent.
The dispute concerned whether conduct amounted to acceptance of contractual terms.
The Court considered Articles 129, 132, 135 and 142 of the UAE Civil Transactions Law.
Article 132 permits an expression of intent to be made:
orally;
in writing;
through customary means;
through an interchange of acts demonstrating mutual consent; or
through another course of conduct where the circumstances leave no doubt that mutual consent has been demonstrated.
The Court held that contractual acceptance does not necessarily require a formal written document. Conduct can demonstrate consent when the circumstances establish mutual intention.
Relevance to machine-mediated consent
This is highly significant.
A machine-mediated transaction can be understood as a technologically expressed form of conduct.
For example:
A company programs an automated purchasing system to accept qualifying offers.
The system's conduct may be evidence of the company's legally attributable consent.
Principle
UAE contract law focuses on demonstrated mutual consent rather than insisting upon a single traditional form of expression.
8. Case Law 2 — Ondina v Olin
[2025] DIFC CFI 046
This case provides an important authority on electronic signatures and contractual modification.
The dispute involved an exchange of emails concerning the amendment of an employment contract.
The Court examined whether the emails were sufficient to satisfy a statutory requirement for an agreement to be in writing and signed.
The Court considered the DIFC Electronic Transactions Law, under which an electronic signature can satisfy a statutory signature requirement. It held that an email ending with the person's name could constitute an electronic signature where the person adopted the name with the intention of signing the relevant electronic record.
Relevance
This demonstrates that:
The legal validity of consent does not depend upon the physical presence of ink on paper.
A technological mechanism can communicate legally effective intention.
Application to AI systems
If an AI platform is configured to place an electronic signature automatically, the critical question becomes:
Was the signature application authorized by the person whom it purports to represent?
The machine's operation alone is not sufficient.
9. Case Law 3 — ICICI Bank Ltd v Bavaguthu Raghuram Shetty
[2022] DIFC CFI 034
This is one of the most important authorities for electronic-signature attribution.
The dispute involved guarantees and other documents containing electronic or copied signatures.
The defendant argued that he had not personally applied some electronic signatures.
The Court examined whether the electronic signatures had been applied by him or with his authorization.
The Court explained that an electronic or copied signature is not inherently invalid. The critical question is whether it was made or authorized by the signatory.
The Court ultimately found, on the evidence, that the relevant electronic signatures were binding where the circumstances demonstrated authorization and recognition of the documents.
The Court of Appeal subsequently emphasized that the relevant question was whether the electronic signatures had been authorized and that this approach was consistent with both DIFC and UAE law.
Importance for machine-mediated consent
This case provides a powerful principle:
Technology does not destroy consent; lack of authorization can.
Therefore, if an AI agent automatically signs or accepts a contract, the central issue may be whether the agent was authorized to act.
10. Case Law 4 — Gideon v Griame LLC
[2016] DIFC SCT 115
The case concerned an employment agreement signed electronically.
The Court considered whether the person who signed the agreement on behalf of the employer had the necessary authority.
The Court applied the concept of apparent authority, examining whether the employer's conduct reasonably caused the other party to believe that the person was authorized to enter into the agreement.
Relevance to machine-mediated consent
This provides an important analogy.
Suppose a company deploys an AI contracting agent on its official website.
The customer reasonably believes:
"This AI agent is authorized to conclude contracts for the company."
The company may face an attribution argument if its own conduct created that reasonable appearance of authority.
Principle
Authority can be assessed by reference to the principal's conduct and the reasonable expectations it creates.
11. Case Law 5 — Naho v Neukirchi
[2024] DIFC SCT 415
This case concerned electronic signatures and the DIFC Electronic Transactions Law.
The Court considered whether information stored electronically, accompanied by a person's adopted name, could constitute an electronic signature.
The statutory definition includes an electronic sound, symbol or process attached to or logically associated with a record and adopted by a person with the intention to sign it.
The case demonstrates that courts examine:
the electronic record;
the identifying symbol or name;
the person's conduct;
the intention to sign;
the connection between the signature and the record.
Relevance
Machine-mediated transactions may involve:
biometric confirmation;
PIN authentication;
one-time passwords;
digital signatures;
platform-generated confirmations.
The legal question is whether the mechanism reliably demonstrates the person's authorization and intent.
12. Case Law 6 — Gate Mena DMCC / Huobi Mena FZE v Tabarak Investment Capital Ltd
[2024] DIFC DEC 002
This Digital Economy Court decision is relevant to modern electronic contracting.
The Court considered the DIFC Contract Law's provisions concerning contract formation.
It noted that:
a contract need not necessarily be in writing;
a contract is concluded by acceptance of an offer;
an offer must be sufficiently definite and demonstrate an intention to be bound;
acceptance may be expressed through a statement or conduct.
Relevance to machine-mediated consent
Digital transactions frequently operate through:
automated trading platforms;
electronic order books;
online marketplaces;
smart contracts;
algorithmic execution.
The decision supports the broader proposition that contractual formation depends upon legally recognizable offer and acceptance rather than upon a particular physical format.
13. Case Law 7 — Barclays Bank PLC v Bavaguthu Raghuram Shetty
[2020] DIFC CFI 061
This case involved questions concerning electronic execution and the authenticity and authority of electronically inserted signatures.
The Court considered allegations that electronic signatures had been inserted without authority and examined the surrounding evidence concerning the execution of the relevant agreement.
The case illustrates a critical distinction:
Electronic appearance of a signature is not necessarily proof of legal consent.
The court may investigate:
who applied it;
who had access;
whether authority existed;
how the document was created;
whether subsequent conduct confirmed the transaction.
Application to AI
An AI agent could potentially apply a signature automatically.
But if:
the agent acted outside its authorization,
the question becomes whether the principal should nevertheless be bound under agency, apparent authority, estoppel, ratification or other applicable principles.
14. Case Law 8 — GFH Capital Ltd v David Lawrence Haigh
[2014] DIFC CFI 020
This case involved electronic signatures, electronic instructions and the use of an electronic signature by another person who had access to it.
The judgment discussed circumstances in which a person's PA had control over the person's electronic signature and could use it under an authorized practice.
Relevance
The case is valuable because it demonstrates that:
Control over a technological authentication mechanism and authorization to use it can become central to attribution.
This is directly relevant to machine-mediated consent.
If a person gives an AI agent authority to use an electronic authentication mechanism, subsequent transactions may potentially be attributed to that person.
15. Case Law 9 — Tarig Mohamed Abdelsalam Abdelrahman v Expresso Telecom Group Ltd
[2021] DIFC CFI 056
The Court considered whether an email communication demonstrated consent to electronic service.
The Court concluded that the particular email did not unequivocally establish consent to service of proceedings by electronic means. The wording was insufficiently clear in the circumstances.
Importance
This case demonstrates the opposite side of machine-mediated consent:
Not every electronic communication necessarily proves consent to the particular legal consequence asserted.
Therefore, the system should distinguish between:
general communication;
acknowledgment;
acceptance;
signature;
authorization;
consent to a specific legal act.
A chatbot saying:
"Thanks, I'll pass that along"
should not automatically be treated as acceptance of a binding contract.
16. The Importance of Article 132 of the UAE Civil Transactions Law
The UAE Civil Transactions Law adopts a flexible concept of expression of intention.
Consent can be demonstrated through:
words;
writing;
customary means;
conduct;
other circumstances demonstrating mutual consent.
The Forbes v Kidd litigation provides a modern judicial discussion of this principle.
This flexibility is particularly suitable for emerging technology.
It allows courts to evaluate new forms of contracting without requiring the legislature to create a separate contractual category for every technological mechanism.
17. Automated Electronic Transactions Under UAE Law
The UAE electronic-transactions framework goes even further than merely recognizing electronic signatures.
It expressly recognizes automated electronic contracting.
The law provides that:
contracts may be formed through automated electronic systems even when there is no direct human intervention at the time of conclusion.
It also contemplates automated systems interacting with human persons where the person knows, or is presumed to know, that the system will automatically conclude or perform the transaction.
This is extremely important for:
algorithmic trading;
automated procurement;
online marketplaces;
smart contracts;
AI agents;
automated renewals;
machine-to-machine commerce.
18. Machine-Mediated Consent vs Human Consent
The following distinction is useful:
| Human consent | Machine-mediated consent |
|---|---|
| Person personally communicates acceptance | System communicates acceptance |
| Intention formed immediately | Intention may have been formed earlier |
| Human directly controls act | System executes programmed instructions |
| Physical/electronic signature possible | Automated authentication possible |
| Attribution usually straightforward | Attribution may require investigation |
| Error usually linked to person | Error may arise from system or programming |
The key issue in the second column is:
Where did the legally relevant intention originate?
19. Pre-Authorization as Consent
A particularly important concept is prior authorization.
Suppose:
A company instructs an AI procurement agent to purchase goods whenever a specified price condition is met.
The company has not personally approved each purchase.
Nevertheless, it may have already expressed consent by creating and authorizing the system.
Therefore:
Prior authorization → automated execution → legally attributable transaction
may be sufficient.
The statutory recognition of automated electronic transactions strongly supports this conceptual model.
20. What Happens When the Machine Makes a Mistake?
Consider:
AI agent is authorized to buy 100 units.
Due to a software error, it purchases 10,000 units.
Several questions arise:
Was the transaction objectively concluded?
Was the AI acting within its apparent authority?
Was the counterparty aware of the error?
Was the error attributable to the principal?
Was there fraud?
Was there a fundamental mistake?
Did the counterparty reasonably rely on the automated system?
Did the principal ratify the transaction afterward?
The answer cannot simply be:
"The machine made a mistake, therefore the contract is void."
Nor can it automatically be:
"The machine acted, therefore the principal is always bound."
The surrounding circumstances matter.
21. AI Agent as Contracting Interface
Modern AI introduces a new category.
An AI agent could potentially:
search for suppliers;
negotiate prices;
select a product;
agree delivery terms;
generate a purchase order;
electronically sign;
transmit the order.
The legal system must then identify:
who owns the agent;
who instructed it;
what authority it possessed;
what restrictions were programmed;
whether the counterparty knew it was dealing with an automated system;
whether the AI exceeded its authority.
The UAE's automated-transaction rules make the basic legal recognition of this model considerably clearer than traditional paper-based contract doctrine.
22. Limits of Machine-Mediated Consent
Automated consent does not eliminate mandatory legal requirements.
Certain transactions may remain subject to:
statutory formality;
notarization;
registration;
identity verification;
regulatory approval;
consumer-protection requirements;
public-policy restrictions;
special rules governing property or family matters.
Historically, Dubai's electronic-transactions legislation also expressly excluded certain categories, including specified family matters, title deeds and certain real-estate transactions.
The precise scope must therefore always be checked against the current federal and sector-specific legislation governing the transaction.
23. Electronic Signature vs Machine Signature
These should not be confused.
Electronic signature
A human adopts an electronic mechanism to sign.
Machine-applied signature
Software applies a signature or authentication mechanism automatically.
The second raises an additional question:
Was the machine authorized to apply the signature?
The reasoning in ICICI Bank v Shetty is especially important here. The relevant inquiry is authorization rather than the mere electronic appearance of the signature.
24. Apparent Authority
Machine-mediated transactions may involve actual authority and apparent authority.
Actual authority
The principal actually authorized the system to conclude contracts.
Example:
"This AI agent may purchase inventory up to AED 100,000."
Apparent authority
The principal's conduct reasonably causes the counterparty to believe that the system has authority.
Example:
The company's official platform repeatedly permits the AI agent to conclude binding orders.
The Gideon v Griame reasoning concerning apparent authority is relevant by analogy.
25. Unauthorized Machine Action
Suppose a hacker takes control of an AI contracting system.
The AI enters a contract worth AED 5 million.
This is materially different from:
The company deliberately programmed the AI to conclude contracts up to AED 5 million.
The first involves unauthorized interference.
The legal analysis may therefore consider:
authentication;
cybersecurity;
authority;
notice;
reliance;
fraud;
negligence;
ratification;
contractual allocation of cyber risk.
The electronic nature of the transaction does not itself answer these questions.
26. Error in AI Negotiation
AI systems may misunderstand:
price;
quantity;
currency;
delivery dates;
contractual conditions;
technical specifications.
A human negotiator who makes an error and an AI agent that makes an error may raise similar contractual questions, but machine-mediated contracting introduces an additional issue:
Was the error caused by the principal's instructions, the system's programming, the vendor, or an external attack?
Attribution therefore becomes essential.
27. Machine-Mediated Consent and Consumer Contracts
Consumer platforms commonly use:
clickwrap agreements;
app-based acceptance;
automated subscriptions;
one-click purchases;
AI customer-service interfaces.
The business should be able to demonstrate:
what terms were presented;
when they were presented;
what the consumer did;
whether the consumer could reasonably understand the transaction;
whether mandatory disclosures were provided.
A system saying:
"User accepted"
is not necessarily sufficient evidence unless the underlying records establish what the user actually accepted.
28. Evidence of Machine-Mediated Consent
Important evidence can include:
transaction logs;
timestamps;
IP records;
authentication records;
device information;
electronic signatures;
system instructions;
API logs;
audit trails;
AI prompts;
model outputs;
authorization rules;
system configuration;
email confirmations.
The more sophisticated the system, the more important reliable audit trails become.
29. Burden of Proof
Where a party denies consent, the court may have to determine:
Did the disputed person actually authorize the transaction?
ICICI Bank v Shetty demonstrates the importance of evidence concerning the authenticity and authorization of electronic signatures.
Evidence may include:
expert analysis;
electronic records;
subsequent conduct;
communications;
system access;
prior transactions;
corporate authorization.
30. Ratification
Even if an automated transaction was initially unauthorized, subsequent conduct may become important.
For example:
An AI agent concludes a transaction outside its authority.
The principal subsequently:
accepts delivery;
pays the invoice;
confirms the order;
performs the agreement.
Such conduct may become relevant to whether the principal subsequently ratified or affirmed the transaction, depending upon the applicable law.
31. Subsequent Conduct as Evidence of Consent
The Forbes v Kidd reasoning demonstrates the importance of conduct in determining mutual consent.
This is especially useful in machine-mediated disputes.
Suppose an AI agent concludes a contract.
The company subsequently:
receives goods;
pays the supplier;
sends a confirmation;
performs the contract for six months.
Those facts may provide powerful evidence concerning whether the company accepted or ratified the transaction.
32. Smart Contracts
Smart contracts present a particularly strong form of machine-mediated consent.
The parties may agree:
"If condition X occurs, system Y automatically transfers asset Z."
The smart contract does not necessarily eliminate the underlying legal contract.
Instead, the technology may provide:
automated execution;
evidence of performance;
predetermined conditions;
reduced need for manual intervention.
The legal question remains whether the underlying arrangement satisfies the applicable contractual and statutory requirements.
33. AI Negotiation and Formation
An AI agent may negotiate:
Seller: AED 100,000.
AI agent: AED 90,000.
Seller: Accepted.
Was a contract formed?
The answer depends on:
whether the AI had authority;
whether the communication constituted an offer;
whether the response constituted acceptance;
whether essential terms were agreed;
whether the parties intended to be bound.
The machine's status as a machine does not itself prevent contract formation.
34. Human Review
For high-value or legally sensitive transactions, human review provides an important safeguard.
Examples:
real-estate transactions;
guarantees;
large loans;
long-term commercial agreements;
employment termination agreements;
regulated financial transactions.
Human review can reduce disputes concerning:
authority;
mistake;
misunderstanding;
fraud;
system malfunction.
However, the absence of human review does not automatically invalidate an automated transaction where the applicable law expressly recognizes automated contracting.
35. Consumer Protection and Unfair Terms
Machine-mediated consent also raises questions concerning whether a consumer genuinely received adequate information.
A company cannot necessarily rely on:
"The user clicked Accept"
if mandatory consumer protections require additional disclosures or prohibit particular contractual terms.
Thus:
technological acceptance ≠ automatic validity of every term.
The method of acceptance and the substantive legality of the contract remain separate questions.
36. Data Protection and Consent
AI systems may also process personal information.
Here, another form of "consent" may be involved:
contractual consent;
data-processing consent;
electronic-signature consent;
marketing consent;
biometric consent.
These concepts should not be conflated.
A person agreeing to a purchase does not necessarily consent to every possible use of their personal information.
37. Machine-Mediated Consent and Identity
Authentication is another critical element.
A system may establish that:
"Account 123 clicked Accept."
But the legal question may be:
"Who controlled Account 123?"
Possible authentication mechanisms include:
password;
OTP;
digital certificate;
electronic signature;
biometric authentication;
hardware token.
The stronger the authentication mechanism, the easier it may be to establish attribution, although no technical mechanism is infallible.
38. Contractual Risk Allocation
Commercial contracts involving AI agents should address:
scope of automated authority;
transaction limits;
approval thresholds;
authentication;
audit trails;
human escalation;
system failures;
cybersecurity;
unauthorized transactions;
liability for vendor failures;
indemnities;
termination.
This can substantially reduce later disputes over whether a machine's action was authorized.
39. Practical Attribution Framework
When determining whether machine-mediated consent is legally valid, a court could examine:
Step 1 — Identify the transaction
What contract or legal act is alleged?
Step 2 — Identify the machine
What system generated the communication?
Step 3 — Identify the principal
Who owned or controlled the system?
Step 4 — Determine authorization
Was the machine authorized to act?
Step 5 — Determine the scope
Did it act within its authorized parameters?
Step 6 — Examine the counterparty
Did the counterparty know or reasonably believe that the system was automated?
Step 7 — Examine the communication
Did it objectively constitute offer or acceptance?
Step 8 — Examine evidence
Are there reliable logs, signatures and authentication records?
Step 9 — Consider error or fraud
Was the result caused by malfunction, hacking, mistake or unauthorized use?
Step 10 — Apply mandatory law
Are there formalities or statutory restrictions?
40. Important Distinction: Validity vs Attribution
These are separate questions.
Validity
Was a legally enforceable contract formed?
Attribution
Can the machine's act be legally attributed to the person or company?
A transaction may satisfy the formal requirements of electronic contracting but still generate an attribution dispute.
For example:
An electronic signature exists, but the signatory claims that a third party applied it without authority.
The issue is not simply whether electronic signatures are legally valid.
The issue is:
Who authorized the signature?
This distinction is central to ICICI Bank v Shetty.
41. Important Distinction: Consent vs Authentication
Authentication establishes:
"This credential/account/signature is associated with X."
Consent establishes:
"X intended to undertake this legal obligation."
They are related but not identical.
A stolen password may authenticate a transaction technologically without establishing genuine authorization.
Similarly, a copied electronic signature may resemble a person's signature without proving that the person authorized its use.
42. Important Distinction: AI Assistance vs AI Agency
AI assistance
The human makes the final decision after receiving AI advice.
Example:
AI recommends Supplier A; manager approves.
The human's decision is relatively straightforward to attribute.
AI agency
The AI itself automatically concludes the transaction within delegated authority.
Example:
AI negotiates and accepts Supplier A's offer.
Attribution becomes more complex.
Fully autonomous machine-to-machine contracting
Two automated systems interact without human intervention.
The UAE electronic-transactions framework expressly contemplates this type of automated contracting.
43. Case-Law Synthesis
The cases collectively demonstrate several principles.
| Case | Main principle | Relevance |
|---|---|---|
| Forbes v Kidd | Consent may be demonstrated through conduct | Automated conduct |
| Ondina v Olin | Electronic communication can satisfy signature requirements | AI/e-signature |
| ICICI Bank v Shetty | Authorization is critical to electronic signatures | Machine attribution |
| Gideon v Griame | Apparent authority can arise from conduct | AI platform authority |
| Naho v Neukirchi | Electronic signature can be created through electronic processes | Digital consent |
| Gate Mena v Tabarak | Offer and acceptance can operate in digital transactions | Automated contracting |
| Barclays v Shetty | Electronic execution requires analysis of authority/authenticity | Unauthorized automation |
| GFH Capital v Haigh | Control and authorized use of electronic signatures matter | Delegated machine authority |
| Tarig v Expresso Telecom | Electronic communication must actually demonstrate the required consent | Limits of implied consent |
44. Six Key Legal Principles
Principle 1 — Electronic form does not invalidate consent
A contract does not become invalid merely because it was concluded electronically.
Principle 2 — Automated contracting can be legally effective
UAE legislation expressly recognizes contracts formed by automated electronic systems without direct human intervention.
Principle 3 — Authorization is fundamental
An electronic signature or machine action must be attributable to the relevant person.
ICICI Bank v Shetty is particularly important here.
Principle 4 — Conduct can demonstrate consent
Forbes v Kidd demonstrates the broad UAE concept of consent through conduct.
Principle 5 — Apparent authority can matter
Gideon v Griame illustrates the importance of the principal's conduct and the reasonable appearance of authority.
Principle 6 — Not every electronic communication constitutes consent
Tarig v Expresso Telecom demonstrates that an electronic communication must actually establish the relevant consent in the circumstances.
45. Possible Defences Against Machine-Mediated Contract Formation
A party may potentially argue:
A. No authorization
The machine acted without authority.
B. System compromise
The system was hacked or credentials were stolen.
C. Fundamental mistake
The machine produced a result materially different from the authorized transaction.
D. Lack of essential terms
The system's communication did not establish a sufficiently definite agreement.
E. Lack of required formality
The transaction was one for which electronic contracting alone was insufficient.
F. Fraud
The apparent automated transaction was manipulated.
G. Lack of attribution
The claimant cannot establish that the disputed system belonged to or was controlled by the alleged contracting party.
46. Liability for AI-Agent Errors
Where an AI agent makes an erroneous decision, the legal responsibility could potentially be distributed among:
the AI owner;
the AI developer;
the platform operator;
the user;
the authentication provider;
the cybersecurity provider.
The correct defendant therefore depends on:
control + authorization + duty + causation + contractual allocation of risk.
47. Future UAE Litigation Issues
Machine-mediated consent is likely to generate disputes involving:
AI contract negotiators;
autonomous procurement;
algorithmic trading;
smart contracts;
automated insurance;
AI-generated employment contracts;
autonomous purchasing;
digital assets;
machine-to-machine commerce;
AI-powered marketplaces;
automated loan agreements.
Future cases may need to determine whether an AI agent has acted:
within actual authority;
within apparent authority;
outside its authority;
because of software malfunction;
because of defective instructions;
because of cybersecurity compromise.
48. Overall UAE Legal Position
The UAE legal framework is relatively clear on the basic validity of automated electronic contracting.
The law does not require a human being to physically click a button at the exact moment a contract is concluded.
Instead, it recognizes:
electronic offer + electronic acceptance + automated systems + legal attribution
as capable of producing enforceable contractual relationships.
The difficult questions arise at the next level:
Was the machine authorized?
Who controlled it?
What was the scope of authority?
Did it exceed its parameters?
Was the counterparty aware of the automated nature of the transaction?
Was the system compromised?
Was the transaction subsequently ratified?
Were mandatory formalities satisfied?
49. Conclusion
Machine-mediated consent is capable of being legally valid under UAE law.
The most important statutory development is the UAE's express recognition of automated electronic transactions, including contracts formed between automated electronic systems without direct human intervention.
The legal theory can therefore be summarized as:
The machine does not need independent legal consciousness; the law can attribute its authorized operation to the person or legal entity that programmed, controlled or authorized it.
The case law reinforces this approach.
Forbes v Kidd demonstrates that consent may be established through conduct.
Ondina v Olin confirms the legal significance of electronic signatures.
ICICI Bank v Shetty emphasizes authorization of electronic signatures.
Gideon v Griame illustrates apparent authority.
Naho v Neukirchi confirms that electronic processes can satisfy signature requirements.
Gate Mena v Tabarak demonstrates the application of ordinary offer-and-acceptance principles to digital transactions.
Barclays v Shetty shows why authenticity and authority must be investigated.
GFH Capital v Haigh illustrates the significance of control over electronic signatures.
Tarig v Expresso Telecom shows that electronic communication must actually establish the required consent.
Therefore, the central UAE civil-law principle is:
Machine mediation changes the method by which consent is expressed and executed; it does not necessarily change the fundamental legal requirements of consent, authority, attribution, contractual formation, and enforceability.
For AI agents in particular, the decisive legal question will often be not "Did the AI consent?", but rather:
"Did the AI act as an authorized technological extension of a person or legal entity, and do the circumstances establish legally attributable consent?"
Quick Revision Points
Machine-mediated consent is technologically expressed or executed consent.
UAE law recognizes electronic offers and acceptances.
UAE law expressly recognizes automated electronic transactions.
Direct human intervention is not always required at the moment of contract formation.
The machine ordinarily does not become an independent legal person.
Attribution to a human or legal entity remains fundamental.
Actual authority and apparent authority may both become relevant.
Electronic signatures are not invalid merely because they are electronic.
Authorization is critical to electronic-signature validity.
Conduct can demonstrate contractual consent.
Not every email or electronic communication constitutes acceptance.
Machine error does not automatically invalidate a contract.
Unauthorized machine action raises separate attribution and cybersecurity questions.
AI-assisted contracting should be distinguished from fully autonomous contracting.
Mandatory statutory formalities can still restrict electronic contracting.
Audit trails and authentication records are important evidence.
Smart contracts are a significant form of machine-mediated performance.
Consumer and data-protection consent should not be confused with contractual consent.
The UAE framework supports technological neutrality in contract formation.
The central question remains authorization + attribution + mutual consent + legal validity.

comments