Generative AI, Stylized Artistic Expression, and Intellectual Property Governance: Copyright, Trademark, and Dataset Liability in AI-Driven Visual Art

Abstract

Generative artificial intelligence has altered the legal understanding of creativity by shifting artistic production from direct human execution to prompt-based, data-dependent image generation.1 The controversy surrounding “Ghibli-style” AI visuals reflects a broader intellectual property problem: modern AI systems do not merely create outputs; they learn from vast collections of protected works, cultural expressions, and recognizable artistic identities.2 This article examines whether AI-generated visual works inspired by a distinctive animation style can attract copyright, infringe protected expression, dilute trademarks, or create liability for users and platforms.3 It further argues that legal analysis should not remain limited to the final image produced by an AI system, but must also extend to the training data, prompting practices, and commercial circulation of such outputs.4 Adopting a doctrinal and comparative method, the article evaluates Indian copyright and trademark law alongside developments in the United States, Japan, the European Union, and China.5 It highlights the distinction between unprotected artistic style and protected expression, the continued relevance of human authorship, the uncertainty surrounding text-anddata mining, and the risks of using brand-specific prompts in commercial contexts. The article also considers cultural misappropriation, platform accountability, digital watermarking, licensing models, and dataset transparency as emerging governance tools. It concludes that India requires a balanced AI-IP framework that protects creators and cultural identity without obstructing legitimate technological innovation.