Protecting Privacy and Copyright Throughout the Data Lifecycle of AI Systems

Abstract

Artificial Intelligence (AI) technology has been evolving rapidly and is now being applied across a wide range of sectors. and being applied in various sectors. While it offers remarkable benefits it also brings considerable challenges, particularly in terms of privacy and copyright. The capabilities of machine learning models come from large-scale datasets, often sourced from the internet, which inherently contain personal and copyrighted materials, leading to complex legal and ethical dilemmas. Despite the transformative potential of AI, the legal and technical landscapes surrounding privacy and copyright protections remain unclear, which has prompted the assessments on the impact of AI on these fundamental rights and the quest for responsible practices that ensure the ethical use of AI. This thesis aims to address the multifaceted challenges that AI technologies pose to privacy and copyright protections while exploring practical solutions for promoting responsible AI development. Through a systematic analysis on the emergent AI technologies and their privacy and copyright implications, this study investigates the complexities of the AI data lifecycle, highlighting legal and technical challenges throughout the process of their development and application. In response to these challenges, the study proposes frameworks and solutions for ethical and compliant data practices, including a responsible web framework for consent management and a Data Bill of Materials scheme to enhance transparency in the AI data supply chain. By identifying key challenges and proposing viable solutions, this thesis provides a comprehensive understanding of privacy and copyright issues of AI from a data lifecycle perspective, and offers practical frameworks and methodologies to guide the responsible development and ethical application of AI systems. Show more