AI Work Under Guardrails : Task Framing for Responsible Delegation in BPM

Abstract

Enterprise AI governance can define principles for accountability, transparency, human oversight, and auditability, but organizations still need to translate these principles into concrete decisions about whether and under what constraints a specific unit of process work may be delegated to artificial intelligence (AI). This research-in-progress paper introduces the Responsibility-Gated Task Frame as a design science artifact for this translation. The artifact extends a descriptive Task Frame with responsibility conditions and supports a task-level decision on the form and constraints of AI-supported delegation. Following a design science research approach, the paper presents the artifact as part of an early design cycle and demonstrates its use through a vignette on AI-assisted compliance assessment based on International Organization for Standardization (ISO) management-system standards. The contribution is a task-level mechanism for operationalizing Responsible Business Process Management (RBPM) and AI governance before AI implementation begins.