<?xml version="1.0" encoding="UTF-8"?>
<reference anchor="I-D.sato-soos-hem" target="https://datatracker.ietf.org/doc/html/draft-sato-soos-hem-03">
   <front>
      <title>The Human Escalation Mechanism (HEM) for Agentic AI Systems</title>
      <author initials="" surname="Sato" fullname="Tom Sato">
         <organization>MyAuberge K.K.</organization>
      </author>
      <date month="May" day="29" year="2026" />
      <abstract>
	 <t>   An AI agent that has been authorized to act autonomously has no
   inherent mechanism to stop itself.  If its mission requires a
   decision that exceeds its authorization, if policy mandates human
   judgment before proceeding, or if the agent itself reaches the
   boundary of its reliable competence, what happens?  Without a
   protocol specifying the answer, one of three failure modes occurs:
   the agent proceeds beyond its authorization and executes actions
   that no human approved; it stalls silently with no notification
   to any principal; or it continues running under a mission that has
   already entered a terminal state, producing actions with no
   legitimate purpose.  In all three cases, the humans responsible
   for the system find out too late.

   This document defines the Human Escalation Mechanism (HEM): a
   normative protocol specifying what a Governance Execution
   Controller (GEC) does when an AI agent session requires human
   judgment before execution may continue.  HEM replaces the three
   failure modes above with a single governed path: the GEC places
   the session into a formally defined HEM_PENDING state, routes a
   structured escalation request to one or more designated human
   principals along an ordered designation chain, enforces a
   prohibition on all state transitions until a human decision is
   received, and processes five defined human decision types.  HEM
   also defines the Policy Rationale Declaration (PRD), which links
   Cedar policies that route to HEM with machine-readable rationale,
   and the Decision Rationale Record (DRR), which captures the human
   principal&#x27;s reasoning for audit and learning purposes.

   For human principals, HEM provides more than oversight: it
   provides better project outcomes, a learning record of both human
   and agent decisions, and a normative channel for injecting real-
   world context into agent execution at the moment it is most
   relevant.  For AI agents, HEM provides a governed recovery path
   that enables higher confidence operation near decision boundaries
   and a shorter, more efficient route to correct outcomes.

   HEM is the entry point to a four-draft governance stack.  The
   Intent Declaration Primitive [I-D.sato-soos-idp] makes HEM
   escalation requests legible to human principals.  The Governance
   Audit Record [I-D.sato-soos-gar] provides the permanent, non-
   suppressible record that oversight occurred and what was decided.
   The Constitutional AI Protocol [I-D.sato-soos-cap] defines the
   prohibition floor below which no HEM decision may operate.

   HEM is enforced by the GEC, not by the agent and not by the
   application layer.  An agent cannot opt out; an application
   cannot suppress it.  This non-bypassability is the source of
   HEM&#x27;s regulatory utility and provides the technical specification
   for human oversight required by EU AI Act Article 14.

	 </t>
      </abstract>
   </front>
   <seriesInfo name="Internet-Draft" value="draft-sato-soos-hem-03" />
   
</reference>
