. Additional leading case laws on automated and electronic contracting

The following cases are particularly useful because they address contract formation through electronic systems, automated transactions, and the allocation of contractual risk. They are not all AI cases, but they provide a foundation for analysing AI agents.

Case 12: Chwee Kin Keong and Others v Digilandmall.com Pte Ltd [2005] 1 SLR(R) 502

Jurisdiction: Singapore

Facts: An online retailer's website displayed printers at a price far below the intended price because of a pricing error. Several buyers placed orders, and the dispute concerned whether enforceable contracts had arisen and whether the buyers could rely on the apparent price.

Legal issue: How ordinary contract-formation and mistake principles apply to transactions concluded through an automated website.

Principle: The Singapore Court of Appeal examined offer, acceptance, electronic contracting, and the doctrine of unilateral mistake. It recognised the relevance of electronic ordering while considering whether the circumstances permitted relief for a serious pricing mistake.

Relevance to AI contract governance: An AI purchasing agent may accept an erroneous price, issue an order, or negotiate a commercially irrational term. The case illustrates that electronic formation does not eliminate ordinary contract doctrines, including mistake, notice, and the parties' knowledge of an obvious error.

Case 13: B2C2 Ltd v Quoine Pte Ltd [2020] SGCA(I) 02

Jurisdiction: Singapore

Facts: Automated trading algorithms on a cryptocurrency exchange entered into trades at unusual prices after a technical change affected the platform's order-matching environment.

Legal issue: How contract and mistake principles apply when algorithmic systems execute transactions without direct human intervention in each trade.

Principle: The Singapore International Commercial Court and, on appeal, the Singapore Court of Appeal considered the platform's contractual arrangements and the doctrine of unilateral mistake in the context of algorithmic trading. The Court of Appeal addressed the relevant state-of-mind inquiry and the contractual allocation of risk, rather than treating the existence of an algorithm as a complete answer.

Relevance to AI contract governance: The case is especially valuable for autonomous AI agents that negotiate or execute transactions. It highlights the need to examine the system's role, the governing terms, the parties' knowledge, and the applicable mistake doctrine. It does not establish a general rule that every unexpected algorithmic result invalidates a contract.

Case 14: Thornton v Shoe Lane Parking Ltd [1971] 2 QB 163

Jurisdiction: England and Wales

Facts: A motorist entered an automatic car park, received a ticket from a machine, and later disputed an exclusion clause displayed or incorporated through the parking process.

Legal issue: When contractual terms are incorporated into a contract formed through an automated transaction and what notice is required for particularly onerous terms.

Principle: The Court of Appeal held that the contract was formed when the motorist entered the car park and the machine issued the ticket. An unusually onerous exclusion term required adequate notice before or at the point of contract formation.

Relevance to AI contract governance: AI agents often accept standard terms through websites, APIs, and automated purchasing workflows. The case illustrates why businesses should present material contractual terms at the proper stage and should not assume that obscure terms automatically bind the other party merely because a machine completed the transaction.

Case 15: Register.com, Inc. v Verio, Inc., 356 F.3d 393 (2d Cir. 2004)

Jurisdiction: United States

Facts: Verio repeatedly accessed information from Register.com's website after receiving notice of terms governing use of the information.

Legal issue: Whether repeated use of a website service with knowledge of the provider's terms could support contractual obligations.

Principle: The Second Circuit considered notice, conduct, and the circumstances in which repeated access to an online service may establish assent to its terms.

Relevance to AI contract governance: An AI agent may repeatedly query a commercial API, access a database, or place orders under stated conditions. The case helps explain why system designers should preserve evidence of notice, assent, and the scope of authorised use. Whether a particular AI agent's actions bind its operator remains a fact-sensitive question.

Case 16: Specht v Netscape Communications Corp., 306 F.3d 17 (2d Cir. 2002)

Jurisdiction: United States

Facts: Users downloaded software from a website where the relevant licence terms were not displayed in a manner that reasonably alerted them before downloading.

Legal issue: Whether the users had assented to contractual terms that were available through a link but not adequately presented.

Principle: The Second Circuit held that the users had not assented to the arbitration clause because the website did not provide reasonably conspicuous notice and an adequate manifestation of assent.

Relevance to AI contract governance: An AI-powered service cannot necessarily bind users to terms that are hidden, inconspicuous, or not reasonably presented. Clickwrap, browsewrap, and automated API terms must be assessed under the applicable jurisdiction's rules on notice and assent.

Case 17: Air Canada v. Moffatt, 2024 BCCRT 149

Jurisdiction: British Columbia, Canada

Facts: A passenger relied on information supplied by Air Canada's website chatbot concerning bereavement fares. The chatbot's information conflicted with the airline's applicable policy, and the passenger sought a remedy.

Legal issue: Whether the airline was responsible for misleading information provided by its customer-service chatbot.

Principle: The Civil Resolution Tribunal held Air Canada responsible for the negligent misrepresentation in the circumstances and rejected the argument that the chatbot should be treated as a separate entity from the airline's website service.

Relevance to AI contract governance: This is a directly relevant AI-chatbot decision. It demonstrates the risk of a business denying responsibility for its own customer-facing automated system. It is a tribunal decision, not a general appellate ruling binding every court, and its application depends on the facts and applicable law.

7. Comparative analysis of AI contract governance

The jurisdictions are not interchangeable. For example, a result under Singapore's doctrine of mistake cannot automatically be applied under Indian contract law, and a US clickwrap decision does not conclusively determine whether an electronic term is incorporated into a contract governed by English law.

8. A practical framework for AI contract governance

Businesses deploying AI agents should adopt a governance framework that connects technical permissions with legally authorised contracting activity.

1. Identify the contracting party

Clearly identify the person or legal entity that is making the offer, accepting terms, placing an order, or undertaking performance. Do not treat the AI system as the contracting party merely because it generated the message.

2. Define the agent's authority

Set clear financial limits, approved counterparties, permitted contract types, and restrictions on warranties, indemnities, exclusivity, and other material terms.

3. Control contract formation

Specify when an offer becomes binding, how acceptance is communicated, which version of the terms applies, and which actions require human approval.

4. Maintain reliable records

Preserve the relevant prompts, approved instructions, transaction logs, contract versions, timestamps, and acceptance messages in proportion to the risks and applicable recordkeeping duties.

5. Plan for errors and disputes

Establish escalation procedures for unusual prices, conflicting terms, unauthorised purchases, model failures, and disputed acceptance. A technical reversal does not necessarily cancel an already formed contract.

6. Allocate risk fairly

Contractual clauses should address system errors, security incidents, vendor responsibilities, warranties, indemnities, and dispute resolution, while respecting mandatory law and limits on excluding liability.

9. Key legal principles

Several principles emerge from comparative AI contract governance.

First, AI can facilitate a contract without becoming a legal person. In many legal systems, the relevant question is whether the actions of the system can be attributed to a contracting party and whether the ordinary requirements for contract formation have been satisfied.

Second, automation does not eliminate consent and authority questions. Businesses must consider whether the system was authorised to make the commitment, whether the parties manifested assent, and whether any applicable rules concerning mistake, misrepresentation, or unfair terms affect enforceability.

Third, unexpected outputs do not automatically invalidate contracts. The result depends on the governing law, the circumstances of formation, the parties' knowledge, and any relevant contractual allocation of risk.

Fourth, the terms governing an AI system are important but not conclusive. A contract may allocate risk between a business and its AI vendor, but that allocation does not necessarily determine the rights of a third-party counterparty or consumer.

Fifth, legal responsibility should be evaluated separately from technical fault. A vendor's software error may be relevant, but a business that deploys the system may still have contractual obligations to its counterparty. Questions of contribution, indemnity, and recourse between the business and vendor are separate matters.

10. Conclusion

Comparative AI contract governance examines how established contract doctrines can accommodate AI systems that draft, negotiate, accept, and perform contractual obligations.

The United States and other common-law jurisdictions have developed substantial experience with electronic contracting and automated transactions. Singapore's Quoine and Digilandmall decisions are particularly important for algorithmic transactions and mistake. The EU's legal framework combines electronic contracting with consumer, data-protection, and AI-specific rules. India's Information Technology Act 2000 and Indian Contract Act 1872 provide a foundation for electronic contracts and contractual obligations, although difficult questions about highly autonomous AI agents may require further judicial clarification or legislation.

The cases discussed above show that contract formation, notice, authority, mistake, attribution, and representations remain central. Air Canada v. Moffatt is particularly relevant to AI chatbot representations, while Specht, Register.com, and Thornton help explain notice and incorporation of terms in automated transactions.

Ultimately, effective AI contract governance requires clear authority, reliable attribution, valid contract formation, fair allocation of risk, and accessible dispute resolution. The law should permit useful automation while ensuring that businesses cannot evade contractual responsibility merely because an AI system carried out the relevant action.