Artificial intelligence has changed how creative content is made — and Indian copyright law simply hasn’t caught up. The Copyright Act of 1957 was written with human creators in mind and is relatively silent when systems like GPT-4, Stable Diffusion, or Mid journey produce entire works independently. Who owns the output? Was training on copyrighted data even legal? The uncertainty is real, and its consequences are growing. This paper works through four questions: whether AI-generated works qualify for protection under Indian law; who holds authorship and ownership rights; when training AI on copyrighted material becomes infringement; and whether Section 52’s fair dealing provisions can realistically handle generative AI. It draws on doctrinal analysis and compares India’s approach against the US, EU, UK, Japan, and Singapore. The gaps are hard to ignore. No data mining exception, no framework for computer-generated authorship, no deepfake legislation — courts are stretching decades-old rules over problems they were never meant to solve. Though the ANI Media case against OpenAI highlights mounting pressure, no single verdict can resolve deep-rooted flaws. A judge’s decision might clarify legal boundaries – yet systemwide issues remain untouched. For India, progress means rethinking copyright with precise guidelines on who made what. Licensing systems for data used in machine learning could follow. Clearer expectations around disclosure might support artists while keeping new ideas flowing. Rules should balance fairness and invention, nothing more. The paper recommends a calibrated statutory framework that preserves protection for demonstrable human creative contribution, clarifies ownership where AI functions as a tool, introduces a limited and transparent data mining/licensing mechanism for training datasets, and strengthens disclosure duties for AI-assisted works. Such reform would better balance creator protection, legal certainty and innovation in India’s emerging AI economy.