The Global Ethics of Foundation Models Trends, Challenges, and Governance Across Disciplines

Abstract

The rapid adoption of foundation models (FMs) such as large language models, multimodal AI systems, and generative AI has raised critical ethical concerns across multiple disciplines. These models, which can execute a variety of tasks with limited oversight, have presented significant risks such as bias amplification, privacy violations, accountability deficiencies, and disparities in access. This paper examines global patterns of ethical application of core models, identifies SIGNIFICANT challenges, and evaluates approaches to governance across academia, industry, and government. A survey was conducted with 185 participants, including AI practitioners, policymakers, academics, and users, to evaluate their awareness, perspectives, and practices concerning the ethics of foundation models. The results indicate a shared agreement on the need for uniform governance structures, improved transparency, and increased interdisciplinary cooperation to support the ethical deployment of AI technology.