Fiduciary Duties as AI Alignment Criteria for Assistants

A new paper on arXiv (2608.02660) proposes that fiduciary theory, drawn from business ethics and legal scholarship, should govern the alignment of advanced AI assistants. The author argues that the developer-user relationship in the user-AI-developer triad is analogous to a fiduciary relationship, given the developer's discretionary control over the system's behavior, memory, and engagement parameters. The paper suggests that the four canonical fiduciary duties—loyalty, care, good faith, and candour—can generate alignment criteria for extended AI assistant deployment. This approach extends beyond traditional moral frameworks like bioethics and virtue ethics, offering a legal and ethical foundation for AI alignment in roles such as advice, decision support, collaboration, learning, emotional support, and companionship. The paper is categorized as a cross-announcement and was published on arXiv.