Outlet Title

Thirty-second Americas Conference on Information Systems, Reno, 2026

Document Type

Conference Proceeding

Publication Date

2026

Abstract

The emergence of Agentic AI systems is reshaping organizational processes through continuous adaptation, autonomous coordination, and real-time operational execution. Unlike traditional artificial intelligence applications focused on discrete decision-making tasks, Agentic AI systems operate continuously and interact dynamically with organizational environments. Existing AI ethics literature has primarily emphasized issues such as fairness, bias, transparency, and accountability within static or decision-centric AI contexts. Similarly, traditional AI alignment research has largely focused on ensuring that AI systems pursue intended goals and remain controllable under specified operational conditions. However, these approaches may be insufficient for explaining how ethical risks evolve with continuously operating autonomous systems. This paper develops a conceptual framework for understanding ethical misalignment in Agentic AI systems.  Ethical misalignment is defined as the gradual deviation between intended ethical objectives and emergent system outcomes resulting from ongoing adaptation and autonomous interaction. Rather than focusing primarily on goal alignment or system control, the study conceptualizes ethical misalignment as a longitudinal organizational phenomenon emerging through continuous adaptation and interaction across adaptive enterprise ecosystems.

This study examines limitations of traditional AI ethics frameworks and proposes a framework linking ethical design, adaptive behavior, emergent interactions, and ethical misalignment outcomes, contributing to an emergent perspective on ethics within autonomous organizational systems.

Download

Share

COinS