1. Introduction

Automated Refund Liability Systems in Thailand concern the civil and commercial liability arising when businesses use automated software, artificial intelligence, payment gateways, e-commerce platforms, banking systems, or algorithmic processes to approve, reject, delay, calculate, or execute customer refunds.

These systems are increasingly important in:

E-commerce and online marketplaces.

Banking and electronic payment services.

Airline and travel cancellations.

Digital subscriptions and recurring payments.

Online food delivery and transportation platforms.

Insurance claim settlements.

Automated billing and payment reconciliation.

AI-based consumer complaint management.

The central legal question is:

Who is legally responsible when an automated refund system incorrectly refuses, delays, underpays, or fails to execute a refund?

Under Thai law, automation does not ordinarily remove the legal responsibility of the business operating the system. A company cannot automatically escape contractual, consumer-protection, or tort liability merely because a computer program made the decision.

An automated refund arrangement must therefore be examined through contract law, consumer protection, electronic transactions, payment regulation, and civil liability.

2. Legal framework governing automated refunds in Thailand

A. Thai Civil and Commercial Code

The Thai Civil and Commercial Code (CCC) provides the principal foundation for private-law refund disputes.

Relevant provisions include:

These provisions are particularly relevant when an automated system fails to return money after a contract has been cancelled, a product has been returned, or a service has not been supplied.

For example, if a consumer purchases a product online and the seller accepts its return but the automated refund system repeatedly rejects the transaction, the seller may remain contractually liable.

B. Consumer Protection Act B.E. 2522 (1979)

The Consumer Protection Act protects consumers against unfair business practices.

Its relevance includes:

Misleading refund representations.

Unfair contractual conditions.

Failure to provide promised refunds.

Misleading automated customer-service decisions.

Unfair treatment of consumers using digital platforms.

Failure to provide appropriate complaint-handling mechanisms.

Businesses cannot necessarily rely on automated policies to override mandatory consumer protections.

C. Consumer Case Procedure Act B.E. 2551 (2008)

This legislation provides procedural protection for consumers bringing disputes against business operators.

Its importance includes:

Consumer access to judicial remedies.

Special procedural treatment of consumer disputes.

Judicial consideration of evidence controlled by businesses.

Potential allocation of evidentiary burdens under the statutory framework.

Compensation and other remedies where legally justified.

This is particularly significant because automated refund systems often generate records that are inaccessible to consumers.

D. Electronic Transactions Act B.E. 2544 (2001)

The Electronic Transactions Act recognizes electronic transactions and electronic records.

Its relevance to automated refunds includes:

Electronic contract formation.

Electronic acceptance and cancellation.

Automated transaction processing.

Electronic records of refund requests.

Reliability and evidentiary value of system-generated records.

An electronic refund confirmation may become important evidence of a company's contractual undertaking.

E. Payment Systems Act B.E. 2560 (2017)

The Payment Systems Act establishes a regulatory framework for payment systems and designated payment services.

It is relevant where refund failures involve regulated payment service providers, electronic money, payment processing, or payment infrastructure.

A distinction must be made between:

The seller's obligation to refund the purchase price.

The payment provider's obligation to process the payment instruction.

The bank's obligations concerning the relevant payment account.

These obligations may overlap, but they are not automatically identical.

3. Legal liability of automated refund systems

Automated refund systems may create several different forms of civil liability.

A. Wrongful automated rejection of refunds

A refund system may reject a legitimate claim because of:

Incorrect transaction identification.

Defective refund algorithms.

Incorrect product-return information.

Failure to recognize a cancellation request.

Incorrect application of refund deadlines.

Inaccurate fraud-detection classifications.

If a customer satisfies the contractual and statutory requirements for a refund, an automated rejection does not necessarily extinguish that entitlement.

The business may be liable for breach of contract under Sections 194 and 215 of the CCC.

B. Delayed refund liability

A company may approve a refund but fail to transfer the money within a reasonable or contractually prescribed period.

Examples include:

Refund approval without payment execution.

Repeated automated processing failures.

Refunds stuck in pending status.

Duplicate verification requirements.

Failure to reconcile payments between a merchant and payment gateway.

Where the delay amounts to non-performance or defective performance, contractual damages may be recoverable.

The company may also be responsible for interest where the applicable legal requirements are satisfied.

C. Incorrect refund calculations

Automated systems may calculate refund amounts incorrectly by:

Deducting unauthorized cancellation charges.

Applying incorrect discounts.

Failing to return delivery charges where legally refundable.

Deducting excessive administrative fees.

Refunding only part of the amount paid.

The legality of deductions depends on the contract, applicable consumer-protection legislation, and the circumstances of the transaction.

D. Double deductions and duplicate transactions

A particularly serious issue arises when a customer is charged twice but the automated system recognizes only one payment.

The consumer may seek recovery of the excess amount through contractual remedies or unjust enrichment under Section 406.

Where the payment provider has independently caused the error, its own contractual or statutory responsibilities must also be examined.

E. AI-based refund fraud detection

Businesses increasingly use machine-learning systems to identify suspicious refund requests.

Potential problems include:

False identification of genuine customers as fraudulent.

Automatic account suspension.

Rejection of legitimate warranty claims.

Inaccurate assessment of returned goods.

Discriminatory or inconsistent refund decisions.

The business should be able to establish the factual and contractual basis for refusing a refund. An internal algorithmic classification is not, by itself, proof that a consumer acted fraudulently.

4. Automated transactions and the Electronic Transactions Act

The Electronic Transactions Act is particularly important because Thai legislation expressly recognizes automated electronic systems.

Section 13/2: Automated electronic contracts

Section 13/2 prevents the validity or enforceability of a contract from being denied merely because an automated electronic data exchange system was used without human intervention.

This has significant consequences for automated refund arrangements.

For example:

A platform automatically confirms cancellation.

A refund system automatically approves a refund.

A computer system automatically generates a refund confirmation.

Two automated systems communicate to process a repayment.

The absence of a human employee at each stage does not, by itself, invalidate the resulting contractual transaction.

Section 11: Electronic evidence

Electronic information cannot be rejected as evidence merely because it is electronic.

Relevant evidence may include:

Refund request timestamps.

Automated approval messages.

Payment gateway records.

System-generated transaction histories.

Customer-service communications.

Algorithmic decision logs.

Electronic cancellation confirmations.

The court considers the reliability, integrity, creation, storage, and communication of the electronic records.

Section 17/1: Input errors

This provision addresses certain errors made by an individual when communicating through another person's automated electronic data exchange system, subject to statutory conditions.

It can be relevant to refund disputes involving mistaken transaction inputs, although it is not a general right to cancel every automated transaction.

These provisions are important because they establish that automated transactions can have legal effect while preserving the need to examine their reliability and the parties' actual obligations.

สพธอ.+2

5. Liability of different participants

The party responsible for a failed refund depends on its role in the transaction.

An important distinction is that a software developer does not automatically become liable for every incorrect decision made by a merchant's refund system. The claimant must establish an applicable legal duty, breach, causation, and legally recoverable loss.

6. Important case laws and judicial principles

Preliminary clarification: Thai Supreme Court decisions specifically concerning AI-driven automated refund systems are not established in the authorities discussed below. The following are genuine Thai Supreme Court decisions dealing with closely related principles of restitution, contractual rescission, unjust enrichment, deposits, and repayment. They are relevant by analogy, not judgments directly deciding automated refund liability.

Case 1: Supreme Court Decision No. 8012/2568

Subject: Unjust enrichment and repayment of money received without legal justification.

The dispute concerned payments made under bank guarantees and the legal consequences of a payment received without an underlying legal entitlement.

The Supreme Court considered Section 406 of the CCC and the circumstances in which a recipient must return money obtained without legal justification.

Legal principle:

A person who receives money without a legally sustainable basis may be required to return it as unjust enrichment.

Application to automated refunds:

Suppose an automated payment system transfers money to a merchant after the underlying transaction has been cancelled and the merchant has no remaining entitlement to retain it.

The principle in this decision supports examining whether the merchant must return the payment under Section 406.

The case also demonstrates the importance of identifying the proper claimant and the legal relationship between the parties.

LawNode

Case 2: Supreme Court Decision No. 10850/2559

Subject: Recovery of money paid by mistake and the good-faith recipient.

The case concerned an erroneous payment of a living allowance to a recipient who was not legally entitled to receive it.

The Court examined unjust enrichment under Section 406 and the protection available under Section 412 where money was received in good faith and had already been spent before repayment was demanded.

Legal principle:

The recovery of money paid without legal justification may depend on the circumstances of receipt, including the recipient's good faith and the statutory rules governing restitution.

Application to automated refunds:

If an automated refund system mistakenly pays a customer twice, the business may seek recovery of the excess payment.

However, the legal analysis should consider whether the payment was genuinely without legal justification and whether statutory protections applicable to a good-faith recipient are relevant.

An automated accounting error does not independently settle the question of repayment.

LawNode

Case 3: Supreme Court Decision No. 10316/2559

Subject: Contractual deposits and unfair contractual terms.

The dispute concerned a deposit paid under an agreement to purchase property.

The Court examined the contractual provision allowing the seller to retain the deposit when the purchaser failed to perform the agreement.

It also considered the Unfair Contract Terms Act B.E. 2540 and the distinction between a deposit and a contractual penalty.

Legal principle:

The legal consequences of retaining a deposit depend on the nature of the payment, the contractual terms, and the applicable statutory rules. A deposit is not automatically treated as a penalty.

Application to automated refunds:

An online platform cannot necessarily retain a customer's payment simply because an automated refund policy classifies the transaction as cancelled or non-refundable.

The legal character of the payment and the enforceability of the relevant contractual condition must first be determined.

LawNode

Case 4: Supreme Court Decision No. 656/2538

Subject: Deposit forfeiture and contractual performance.

The case concerned a land purchase agreement in which the purchaser paid a deposit and subsequently failed to complete the purchase.

The Supreme Court distinguished a deposit under Sections 377–378 from a penalty under Sections 379–383.

The Court held that the deposit could be forfeited under the applicable contractual and statutory conditions.

Legal principle:

A payment described as a deposit must be examined according to its legal nature and the applicable rules rather than simply being treated as an ordinary refundable payment.

Application to automated refunds:

An automated refund system must distinguish between:

Refundable advance payments.

Legally forfeitable deposits.

Contractual cancellation charges.

Payments that must be returned after rescission.

A system that treats every cancelled payment as automatically refundable, or every deposit as automatically non-refundable, risks producing legally incorrect outcomes.

ThaiDeka

Case 5: Supreme Court Decision No. 918/2558

Subject: Restitution following contractual rescission.

The dispute involved a contractual deposit and additional payments made after an agreement.

The Supreme Court applied Section 391, requiring the parties to restore their previous positions following rescission.

The decision recognized the obligation to return money received under the rescinded agreement, together with applicable interest.

Legal principle:

Rescission creates reciprocal restitution obligations, subject to the statutory rules.

Application to automated refunds:

Where an online purchase or service contract has been validly rescinded, an automated refund system should calculate the repayment by reference to the parties' restitution obligations.

A system cannot treat a validly rescinded transaction as an ordinary completed sale simply because its database still records the transaction as completed.

LawNode+1

Case 6: Supreme Court Decision No. 6967/2538

Subject: Return of deposits after termination of a purchase agreement.

The case concerned a terminated agreement to purchase property and the obligation to return deposits.

The Supreme Court considered the consequences of contractual termination and the requirement under Section 391 that the parties restore their previous positions.

Legal principle:

Where the relevant legal conditions for rescission are satisfied, money received under the contract may have to be returned.

Application to automated refunds:

A platform that has accepted a cancellation and is legally required to restore the customer's payment cannot ordinarily avoid that obligation by relying on an internal automated status that continues to classify the transaction as active.

LawNode+1

Case 7: Supreme Court Decision No. 6542/2567

Subject: Contractual rescission and restoration of the parties' original positions.

This case concerned a construction agreement and the legal consequences of contractual termination.

The Supreme Court considered Section 391 and the obligation to restore the parties to their previous positions. It also addressed situations in which completed work cannot simply be returned.

Legal principle:

Restitution after rescission must take account of the nature of the performance already provided. Where restoration in kind is impossible, monetary compensation may be required under the statutory framework.

Application to automated refunds:

This principle is relevant to digital subscriptions, travel bookings, and online services.

For example, where a customer cancels a service after partial performance, an automated refund system should distinguish between the refundable amount and any legally justified payment for services already provided.

The refund calculation should reflect the actual contractual and statutory position rather than a rigid algorithmic formula.

LawNode

Case 8: Supreme Court Decision No. 16093/2557

Subject: Unfair contract terms and restitution after termination.

This decision considered the Unfair Contract Terms Act B.E. 2540 and Section 391 of the CCC.

The Court addressed the legal consequences of termination and the restoration of the parties' previous positions.

Legal principle:

Contractual provisions must be examined against the statutory framework governing unfair terms and contractual termination.

Application to automated refunds:

An e-commerce business should not assume that every automated cancellation or refund restriction is enforceable.

For example, a standard-form condition stating that all payments are non-refundable may require examination where the business itself has failed to provide the promised service or where mandatory consumer protections apply.

LawNode

7. Consumer protection and unfair automated refund policies

The Consumer Protection Act B.E. 2522 and the Unfair Contract Terms Act B.E. 2540 are important safeguards.

Automated refund systems should not be designed to defeat consumer rights through standard-form contractual conditions.

Examples of potentially problematic practices include:

1. Absolute non-refund policies

A business may advertise that payments are never refundable, regardless of whether the business performs its contractual obligations.

Such a provision must be examined against applicable consumer legislation and contract law.

2. Excessive cancellation deductions

A company may automatically deduct an unreasonable amount from the customer's refund.

The enforceability of such deductions depends on the contractual arrangement and relevant statutory provisions.

3. Artificial refund delays

A business may repeatedly change a refund's status from pending to processing without actually executing payment.

If this amounts to contractual non-performance, the business may incur liability.

4. Unexplained automated rejection

A platform may reject a valid refund without providing a meaningful explanation or a workable complaint process.

This can raise consumer-protection and contractual issues, particularly where the automated decision conflicts with the business's actual obligations.

8. Evidence and burden of proof in automated refund disputes

Evidence is especially important because the company generally controls the technical records of its refund system.

Important categories of evidence

Under the Consumer Case Procedure Act B.E. 2551, special evidentiary rules may apply to consumer cases, including provisions concerning burdens of proof.

A business should not assume that a consumer must independently obtain technical evidence that is exclusively controlled by the business.

9. Civil remedies available to consumers

Depending on the facts and legal basis of the claim, consumers may pursue the following remedies.

A consumer may also pursue an appropriate complaint through the Office of the Consumer Protection Board (OCPB) or the relevant regulator, depending on the nature of the business and dispute.

Where a regulated payment service is involved, the Bank of Thailand's supervisory framework may also become relevant.

bot.or.th+1

10. Practical illustrations

Scenario A: Automated refund rejection after product return

A consumer purchases an electronic device for 25,000 baht.

The consumer returns the product in accordance with the applicable return policy. The seller accepts the returned device, but the automated refund system repeatedly rejects the repayment request.

Legal analysis:

The consumer may establish the contractual refund entitlement.

Section 215 may support a claim for damages arising from non-performance.

Section 391 may apply if the contract has been validly rescinded.

The seller cannot necessarily avoid liability by blaming its software.

Scenario B: Duplicate automated refund

A consumer purchases a product for 8,000 baht.

The automated system mistakenly refunds 16,000 baht.

Legal analysis:

The excess 8,000 baht may constitute unjust enrichment.

Section 406 may provide the basis for recovery.

Section 412 may become relevant to the consequences of good-faith receipt.

The circumstances surrounding the payment and the recipient's knowledge must be examined.

Scenario C: AI-based rejection of an airline refund

A passenger cancels a ticket under circumstances in which a refund is contractually or legally due.

The airline's AI system automatically rejects the claim because its classification algorithm treats the cancellation as a voluntary cancellation outside the refund conditions.

Legal analysis:

The contractual terms and applicable aviation and consumer rules must be examined.

An erroneous AI classification does not determine the passenger's legal entitlement.

The airline may be responsible for its own failure to provide a legally required refund.

Scenario D: Payment gateway failure

A merchant approves a refund of 12,000 baht, but the payment gateway fails to transfer the money.

Legal analysis:

The court would need to examine the separate obligations of the merchant and payment provider.

Relevant questions include whether the merchant properly initiated the refund, whether the provider accepted the payment instruction, and which party caused the failure.

The seller's refund obligation and the payment provider's processing responsibility should not automatically be treated as the same obligation.

11. Challenges in regulating automated refund liability

Thailand faces several important legal and technological challenges.

A. Algorithmic accountability

Businesses must be able to investigate incorrect automated decisions and establish why refunds were approved or rejected.

B. Lack of transparency

Consumers may not understand why their refund applications have been rejected.

C. Cross-border transactions

International e-commerce transactions create questions concerning applicable law, jurisdiction, payment intermediaries, and enforcement.

D. Data protection

Refund systems process personal information, bank details, transaction histories, and identity information.

The Personal Data Protection Act B.E. 2562 (2019) may apply to the collection and processing of this information.

E. Cybersecurity and payment fraud

Businesses must distinguish genuine technical errors from unauthorized transactions and fraudulent refund requests.

F. Multiple-party responsibility

A single refund transaction may involve the merchant, platform, payment gateway, bank, and software provider. Determining which party caused the loss can be difficult.

12. Suggested legal and regulatory reforms in Thailand

Thailand could strengthen automated refund governance through the following measures:

1. Mandatory refund transparency: Businesses should provide clear refund eligibility criteria and reasons for rejection.

2. Human review mechanisms: Consumers should have access to meaningful review when automated systems reject disputed refunds.

3. Refund processing deadlines: Appropriate sector-specific deadlines could reduce unreasonable delays.

4. Automated audit trails: Businesses should preserve reliable records of refund approvals, rejections, and payment execution.

5. Algorithmic accountability: Businesses should periodically test automated systems for systematic errors and unfair outcomes.

6. Clear allocation of liability: Regulations and contracts should distinguish the responsibilities of merchants, platforms, and payment service providers.

7. Effective complaint procedures: Consumers should have accessible mechanisms for challenging erroneous refund decisions.

8. Cross-border cooperation: Consumer and financial regulators should strengthen cooperation in international digital-payment disputes.

13. Overall legal assessment

Established legal framework

Automated refund liability in Thailand is governed by existing principles of contractual obligations, rescission, restitution, unjust enrichment, tort liability, electronic transactions, and consumer protection.

The most important statutory provisions are:

CCC Sections 194, 213 and 215: Performance and contractual liability.

CCC Sections 386 and 391: Rescission and restitution.

CCC Section 406: Unjust enrichment.

CCC Sections 420 and 438: Tort liability and damages.

Electronic Transactions Act Sections 11 and 13/2: Electronic evidence and automated contracts.

Consumer Protection Act B.E. 2522.

Consumer Case Procedure Act B.E. 2551.

Payment Systems Act B.E. 2560.

The eight Supreme Court decisions discussed above provide relevant judicial principles concerning repayment, contractual termination, restitution, and unfair contractual arrangements. They should not be mistaken for direct precedents on AI refund algorithms.

Conclusion: The fundamental principle is that automation does not eliminate civil responsibility. Where a business has a legal obligation to refund money, the use of an automated system does not, by itself, extinguish that obligation. Thai courts can examine the underlying contract, the conduct of the parties, the electronic evidence, the applicable consumer protections, and the cause of the refund failure when determining liability.

Research qualification: This is a general academic legal analysis, not a verified exhaustive survey of Thai Supreme Court judgments. The cited cases are analogous authorities, and the exact application of each provision depends on the facts and the authoritative Thai-language statutory text.