Roboterarme an einer Fertigungslinie in einer hellen Fabrik

Projekt

Artificial Intelligence and Patent Jurisprudence in India: Navigating Inventorship, Section 3(k) Dilemmas, and the Roadmap for Viksit Bharat @2047

Abstract Patents Act drafted in 1970 for a world of mechanical assemblies and chemical formulations and then has to stretch them, sometimes uncomfortably, to fit technologies nobody anticipated. Artificial intelligence is the latest and perhaps the most demanding of these stretches. It is central to the country’s Viks…

Abstract Patents Act drafted in 1970 for a world of mechanical assemblies and chemical formulations and then has to stretch them, sometimes uncomfortably, to fit technologies nobody anticipated. Artificial intelligence is the latest and perhaps the most demanding of these stretches. It is central to the country’s Viksit Bharat @2047 ambitions, driving growth in pharmaceuticals, semiconductor design, fintech, and advanced manufacturing, yet it sits awkwardly within a patent regime built on the presumption that an inventor is a natural person who can be named, questioned, and held accountable. This paper works through three points of friction between AI and Indian patent law, testing each against how comparable jurisdictions the United States, the United Kingdom, and the European Patent Office have responded. The first is inventorship itself: whether a system such as DABUS can ever satisfy Sections 2(1)(y) and 6 of the Patents Act, 1970, and what happens to applications where a human contributor is named largely as a matter of convenience rather than genuine authorship. The second is the Section 3(k) exclusion for computer programmes per se and algorithms, read alongside the Delhi High Court’s technical-effect reasoning in Ferid Allani and its later application to AI-adjacent claims. The third concerns disclosure: what Section 10(4) can realistically demand of a model whose internal decision pathway is, by design, not fully interpretable even to the people who trained it. Rather than treat these as abstract doctrinal puzzles, the paper closes with a set of concrete, implementable proposals clarified guidance on human-contribution thresholds, an updated CRI examination manual with AI-specific illustrations, a Budapest-Treaty-style depository for training artefacts, and time-limited sui generis right for genuinely autonomous outputs. The underlying argument is a modest one: India does not need to choose between protecting human ingenuity and encouraging AI-driven research. It needs sharper definitions, and it needs them before the volume of AI-related filings makes ad hoc case-by-case resolution unworkable.