The EFA Non-Delegation Playbook for Agentic AI V1.2

Version 1.2 replaced V1 10.5281/zenodo.18390725 This playbook operationalizes Ethical Functionality Without Agency (EFA) for agentic AI deployments, providing a deployable guide to maintain human moral authorship in high-speed systems. It includes domain classification (R/M/H), the Human Ethical Agent of Record (HEAR) doctrine, runtime enforcement concepts, Ring of Fire (RoF) for multi-model oversight, bias mitigation requirements, standards crosswalks, and a getting-started checklist. Designed for governance leads and CISOs, it ensures AI remains a tool, not an agent, while addressing 2026 risks like latency gaps and unowned harms.