Introduction
The deployment of artificial intelligence in commercial transactions has advanced at a pace that outstrips the doctrinal assumptions embedded in contract law. When an AI system negotiates terms, generates an offer, and executes a binding commitment without human intervention at each decision node, the classical architecture of the Indian Contract Act 1872 begins to show its structural seams. The statute was drafted in an era when every contractual act presupposed a human mind behind it, and its foundational requirements of free and informed consent, the legal capacity of parties, and the meeting of minds between offer and acceptance were conceived with natural or juristic persons squarely in view. The emergence of autonomous AI systems capable of making commercially significant decisions forces Indian contract law into a confrontation with questions it was never designed to answer.
This article examines whether AI systems can be contracting parties under Indian law, traces the doctrinal inadequacies of the current framework, and situates the Indian position within a comparative and international context to propose a path toward legislative and judicial reform.
Legal Framework
The Indian Contract Act 1872 establishes the foundational requirements of a valid contract in Section 10: the parties must be competent, the agreement must be supported by consideration, it must arise from free consent, and it must not be for an unlawful object. Competence, addressed in Section 11, requires that a party must have attained the age of majority, be of sound mind, and not be disqualified from contracting by any law to which they are subject. This framework contemplates human beings and entities treated as persons in law, such as companies incorporated under the Companies Act 2013 or partnerships registered under the Limited Liability Partnership Act 2008.
AI systems, irrespective of their sophistication, possess no legal personality under Indian law. They are not natural persons. They are not juristic persons. They cannot hold rights, bear obligations, or appear before courts in their own name. The consequence is a fundamental gap: when an AI system concludes a transaction autonomously, the contract ostensibly exists but the doctrinal architecture cannot cleanly assign legal responsibility for it.
The principal-agent framework offers a partial but imperfect solution. Section 182 of the Indian Contract Act defines an agent as a person employed to do any act for another or to represent another in dealings with third parties. The critical word is “person.” An AI system cannot be an agent under Indian law because it lacks legal personhood. What exists instead is a more complex arrangement in which the AI system is a sophisticated tool deployed by a principal, and the question becomes whether the principal can be bound by the acts of that tool under a doctrine of apparent or actual authority.
The Information Technology Act 2000, through Section 11A introduced by the 2008 Amendment, acknowledges the validity of contracts formed through automated electronic processes. Section 10A provides that contracts formed through electronic means shall not be deemed unenforceable solely because no human was involved at the point of formation. This statutory recognition was a significant step, but it addressed automated e-commerce transactions rather than genuinely autonomous AI decisionmaking. The distinction between an automated process that executes predetermined rules and an AI system that exercises judgment by learning from data and generating novel outputs is one that the Information Technology Act does not draw.
Judicial Developments
Indian courts have not yet directly addressed the question of AI-generated contracts, but a body of case law on automated and electronic contracts offers useful analogies. In Trimex International FZE v. Vedanta Aluminium Ltd (2010), the Supreme Court upheld a contract formed through email exchanges, affirming that the formality of a signed paper document is not necessary for a binding agreement. The court’s reasoning emphasised the conduct of parties and the intention discernible from their communications, an approach that is useful but insufficient when the “conduct” is the output of an algorithm.
The problem of machine-generated contracts was alluded to but not resolved in Societe Generale (India Branch) v. Standard Chartered Bank (Bombay High Court, 2021), where automated algorithmic trading instructions were at issue in a dispute about foreign exchange settlement. The court held that the bank was bound by instructions generated through its automated systems, treating the system’s output as the act of the bank. This is an important precedent but operates at the level of attributing machine conduct to the deploying entity, not at the level of resolving whether the machine itself has any contracting capacity.
The National Company Law Tribunal, in Byju’s restructuring proceedings in 2024, had occasion to note the complexity of digital subscription agreements generated through algorithmic personalisation, where the offer presented to each consumer was individually tailored by the platform’s AI. The NCLT’s observations, though obiter, suggested that where the offer itself is AI-generated and the terms are not standardised, the question of whether a genuine meeting of minds occurred is not trivial.
Contemporary Issues and Analysis
The doctrinal difficulty with AI-generated contracts can be analysed under three distinct categories, each presenting its own challenge to established law.
The first is the offer-acceptance problem. A valid contract requires that an offer be communicated by a person with the intention to be bound, and that the acceptance mirror the terms of the offer. When an AI system generates an offer based on its assessment of market conditions, the counterparty’s profile, and real-time data, the question of whose intention is being expressed becomes genuinely complex. The intention is not that of the programmer who designed the system, since the programmer could not have foreseen the specific transaction. It is not that of the deploying company in the sense of a particular human officer who authorised this specific offer. The AI system’s output represents an emergent decision not reducible to a specific human act of will.
The second is the capacity problem. Section 11’s requirements map onto human attributes. An AI system cannot be “of sound mind” in any meaningful sense because it does not have a mind. Extending capacity to AI systems would require either a legal fiction of the sort that treats companies as persons, or an entirely new category of legal entity. India has created no such category, and the debate remains largely academic.
The third is the consideration problem. While consideration need not move from the promisee, it must be something of value in the eyes of law. In fully automated supply-chain contracting where AI systems on both sides negotiate and conclude agreements, the consideration flows between the deploying entities, but the acts of giving and receiving are performed by machines. The courts have not yet been asked to determine whether this creates any defect in the consideration requirement, but the theoretical difficulty is real.
Algorithmic trading presents perhaps the sharpest version of these problems. High-frequency trading systems operating on Indian exchanges execute thousands of contracts per second, each a binding sale or purchase of securities. SEBI’s Algorithmic Trading Regulations, last comprehensively updated through circulars in 2021 and 2022, impose obligations on the exchange member deploying the algorithm but say nothing about the contractual validity of individual machine-generated transactions. The assumption is that the member is always the contracting party, with the algorithm as an undisclosed agent. This assumption holds for current systems but may fracture as AI systems become more autonomous and begin operating across multiple principals simultaneously.
Comparative and International Perspective
The United Nations Commission on International Trade Law (UNCITRAL) has grappled with this issue through its Model Law on Electronic Commerce and, more recently, through its work on automated contracting under the Electronic Communications Convention 2005. Article 12 of the Convention addresses contracts formed by automated message systems and provides that the absence of human intervention does not affect the validity or enforceability of the contract, attributing the contract to the party on whose behalf the system operated. This is a conservative attribution-based approach that avoids recognising AI agency but solves the practical problem of enforcement.
The UK Law Commission’s 2021 Consultation Paper on Automated Vehicles touched on automated contracting and concluded that English common law could accommodate contracts formed by automated processes through the concept of deemed authority. However, the Commission flagged that genuinely autonomous AI systems, where the machine’s decision cannot be traced back to human instructions, create attribution difficulties that common law tools may not resolve. The Commission recommended legislative review, and the Law Commission’s subsequent 2023 report on digital assets and smart contracts continued to recommend targeted statutory intervention.
Singapore has been the most proactive common-law jurisdiction in this regard. The Electronic Transactions Act (Cap. 88) was amended in 2021 to expressly provide that contracts formed through automated systems are valid and binding on the person on whose behalf the system operates, with no requirement for human review at the moment of formation. The Singapore approach establishes clear attribution without resolving the deeper question of AI legal personality, which it defers to future legislative development.
The EU’s AI Act, which entered into force in 2024, does not directly address contractual capacity but creates a risk-based framework for AI systems used in high-stakes commercial contexts, including systems that make or substantially influence autonomous commercial decisions. The EU approach is regulatory rather than private-law based, but its categorisation of high-risk AI applications will inevitably influence how AI-generated contracts in EU jurisdictions are scrutinised for compliance.
Practical and Policy Implications
The practical implications of doctrinal uncertainty in this area are significant for Indian commerce. India’s e-commerce market, valued at over $100 billion in 2025, relies heavily on automated contracting for everything from consumer purchases to seller onboarding agreements. Supply chains in the automotive and pharmaceutical sectors increasingly use AI-powered procurement systems that generate purchase orders, negotiate delivery schedules, and manage contract variations without human sign-off at each step. If the contracts generated by these systems were successfully challenged on the ground that no valid offer or acceptance occurred, or that the deploying company could not be shown to have had contractual intent, the disruption to commercial life would be severe.
For algorithmic trading, the stakes are particularly high. A successful challenge to the validity of machine-generated securities contracts would create systemic risk in Indian capital markets. SEBI’s implicit assumption that all machine-generated trades are valid contracts attributable to the member should be made explicit through regulatory guidance or legislative amendment.
The MSME sector faces a different kind of risk. Small suppliers entering into AI-generated procurement contracts with large buyers are often unaware that the terms they are accepting were generated by the buyer’s AI system rather than negotiated by a human counterpart. The information asymmetry is significant, and the doctrine of unconscionability under Section 16 of the Indian Contract Act (undue influence) is not well-suited to address it.
Suggestions and Reforms
Several reforms deserve serious consideration. First, India should amend the Information Technology Act 2000 to add a dedicated provision on AI-generated contracts, building on Section 10A. The provision should establish a clear attribution rule: a contract generated by an AI system is deemed to be the contract of the person or entity who deployed and operates that system, provided that deployment was authorised. This rule should apply regardless of whether a human reviewed and approved each specific transaction.
Second, SEBI should issue explicit regulatory guidance confirming that contracts generated by algorithmic and AI trading systems are valid and binding on the member, and establishing what technical and audit-trail requirements must be met to support that attribution.
Third, India should consider establishing a legal personality framework for advanced AI systems deployed in commercial contexts, drawing on the concept of a registered AI agent with limited liability exposure. This would require legislative innovation but would resolve the deeper doctrinal problem rather than merely papering over it.
Fourth, the Ministry of Commerce should engage with UNCITRAL’s ongoing work on AI and digital commerce to ensure that any international instruments developed in this space are compatible with Indian legislative developments.
Conclusion
Artificial intelligence as a contracting party is no longer a theoretical curiosity. It is a practical reality that Indian courts, regulators, and legislators are beginning to encounter. The Indian Contract Act 1872 was drafted without any conception of autonomous machine agency, and its requirements of capacity, consent, and intention all assume a human mind. The Information Technology Act provides partial cover for automated contracting but does not address genuinely autonomous AI decisionmaking. As India’s digital economy matures and AI systems take on increasingly significant commercial roles, the gap between legal doctrine and commercial reality will become harder to ignore. Legislative reform, grounded in a clear attribution principle and accompanied by a regulatory framework for AI deployed in high-stakes commercial contexts, is the most pragmatic and legally coherent path forward.
