Training Artificial Intelligence on Copyrighted Works in Pakistan: A Doctrinal Analysis and Proposal for a Comprehensive Legal and Policy Framework

Abstract

The rapid advancement of artificial intelligence (AI) has transformed the creation, dissemination, and utilization of copyrighted works, giving rise to complex legal challenges concerning the use of protected content for AI training. In Pakistan, the Copyright Ordinance, 1962 does not expressly regulate the extraction and use of copyrighted materials for machine learning purposes, resulting in significant legal uncertainty for copyright owners, AI developers, and policymakers. This study employs a qualitative doctrinal research methodology by critically examining Pakistan's copyright framework, relevant international copyright instruments, comparative legal developments, and emerging judicial and policy approaches from selected jurisdictions. The research analyzes the compatibility of existing copyright principles with AI training practices, evaluates the adequacy of available exceptions and limitations, and identifies the legal and policy gaps that hinder innovation while compromising the protection of authors' rights. Based on this analysis, the study proposes a comprehensive legal and policy framework tailored to Pakistan that seeks to balance technological innovation, public interest, and effective copyright protection. The proposed framework aims to provide legal certainty, promote responsible AI development, and align Pakistan's copyright regime with evolving international standards in the digital era.