Who Is Responsible When AI Shapes Corporate Decisions? Director Duties and Corporate Accountability under Indian Company Law

Abstract

A company is deciding whether to acquire another business. An artificial intelligence system examines financial records, market data and compliance risks, and recommends that the deal should proceed. The directors approve it. Months later, the acquisition causes serious loss. The board signed the resolution, but the model shaped the information and the choice. Who was really responsible? This article examines that question through section 166 of the Companies Act, 2013. It argues that directors may use AI, just as they use lawyers, auditors and other advisers, but they must not allow assistance to become substituted judgment. The relevant legal question is not whether the model was perfect. It is whether the directors understood its purpose, considered its limits, questioned important assumptions and retained the practical ability to reject its recommendation. The article distinguishes routine assistance, decision support and decision substitution. It then explains how opacity, automation bias, defective data and fragmented vendor arrangements may weaken accountability. Drawing on Indian company law and comparative principles reflected in the European Union Artificial Intelligence Act and Indian financial regulation, the article proposes five safeguards: disclosure of material AI use, board-level competence, governance-level explanation, records and continuing model review, and non-delegable human responsibility. Corporate governance may become algorithmically informed, but it cannot become algorithmically unaccountable.