When AI Becomes a Business Actor: Closing the Accountability Gap in Agentic Enterprises

Abstract

Artificial intelligence is moving beyond prediction, content generation, and decision support into systems that plan, select tools, communicate with external parties, and execute multi-step business processes with limited supervision. This transition creates the agentic enterprise, in which AI systems function as operational business actors even though they remain incapable of bearing moral blame, legal duties, or institutional responsibility. The resulting accountability gap appears when organizations delegate real authority to AI but cannot identify who authorized an action, which data and rules shaped it, who was expected to intervene, or how an affected person can obtain correction. This conceptual article examines that gap through business ethics, contemporary AI governance, organizational responsibility, and faith-informed public reason. It distinguishes functional agency from moral and legal agency and argues that responsibility must remain with identifiable human principals and deploying institutions. The article proposes an Enterprise Agent Accountability Framework with seven connected controls: named human ownership, bounded delegation, agent identity and authorization, traceable action records, risk-proportionate human review, stakeholder notice and appeal, and remedy with continuous learning. It also introduces a five-level autonomy model for matching controls to business risk and applies the framework to hiring, pricing, customer service, procurement, finance, marketing, and supply-chain operations. The study contributes an interdisciplinary governance model that links technical controls with dignity, justice, transparency, responsibility, and public benefit. It concludes that enterprises may delegate tasks and limited authority to AI agents, but they may not delegate away accountability.