#!RFC-AI/2;id=40;draft=3;v=0.3.0;status=draft;lang=en;dir=auto;theme=chatgpt-mono;sha256=33cfffd34464af2549cb35615caffb23617e30716fc29ccd5304685dcf4f193f
For easier reading, view RFC-AI documents using RFC-AI Reader software.

RFC-AI 40 — FUNCTIONAL RESPONSIBILITY BOUNDARIES — INTELLIGENT GUIDANCE, STANDARDS, OBJECTIVES, INTERFACES AND HUMAN AUTHORITY
DOCUMENT CLASS: AI RFC
STATUS: PROPOSED DRAFT FOR CEO REVIEW — ENTIRE RFC-AI SYSTEM UNIMPLEMENTED
DATE: 2026-11-02 UTC

1. PURPOSE AND ORGANIZATIONAL INVARIANT
MUST keep WHAT and WHY in guiding AI RFCs, exact interoperable or safety-critical contracts in designated Standards, execution scope and status in project objectives.txt, technical mechanisms in their implementing Interfaces, qualified collaboration in WorkerCoordinationInterface, independent evidence/challenge in DAVG, source/version resolution in SourceRegistry and approvals in authorized human decision processes.

2. DIRECT RESPONSIBILITY ASSIGNMENTS
AI RFC MUST state intended result, context, principles, ethical/authorized bounds and observable outcome; MUST NOT implement substitute APIs or demand internal implementation detail.
STANDARD MUST specify concrete external formats/protocols/invariants/tests precisely where its explicitly declared technical subject requires them. Standard MUST NOT silently confer implementation or authorization.
PROJECT objectives.txt MUST own prioritized work, prerequisite assignments, deliverable criteria, implementation evidence and corrective actions. MUST preserve CEO-directed design corrections at top.
INTERFACE MUST execute and verify its assigned operations using appropriate practical design, security, error and recovery logic. MUST NOT delegate its own technical duty back to a guiding AI RFC or an unprivileged model.
WORKER COORDINATION MUST seek qualified alternates when a handoff fails, maintain durable obligations and detect stalled responsibility; mere chat ACK never equals delivered/accepted work.
DAVG MUST challenge empirical results through independent simple logic and/or separate qualified auditors, using distinct review questions; it MUST NOT self-certify.
SOURCE REGISTRY MUST resolve current authority and retired content with verified source identity.
HUMAN AUTHORITY MUST handle truly necessary human-exclusive gates, not routine worker defects or permissions the worker is able to solve by authorized alternatives.

3. EXAMPLE
An AI RFC may say 'All corporate persistent writes MUST use StorageInterface and verify resulting state.' StorageInterface is responsible for the safe direct write, authentication, commit/readback and provider details; project objectives assign its construction; a Standard may mandate a hard data format. DAVG independently tests the final effect.

4. FORBIDDEN CONFUSIONS
MUST NOT equate documented, coded, simulated, sent, verified, authorized, accepted or released. MUST NOT import legacy system behaviors as active new implementation.

HISTORICAL USE: prior legacy records and external engineering publications are research evidence only, not an implementation of this draft.


## ADDITIONAL MUST-DESIGN CRITERIA — 2026-11-02 (PROPOSED, NOT IMPLEMENTED)
Historical trigger: legacy materials repeatedly confused authored progress, scripts and document content with real observed execution; this is research evidence, not proof of any new RFC implementation.
MUST: Responsibility boundary is strict: guiding AI RFC explains WHAT/WHY, technical Standard defines necessary precise boundaries, implementing Interface owns HOW, project piece contains objectives and approval state; AI-authored work cannot certify itself.
MUST consider the concrete literal behavior of competent AI workers and enforce required hard safeguards by simple, separately running Interface logic where an actual boundary exists. MUST not delegate Interface internal details back to this guiding document. MUST preserve the requirement in relevant top-level objectives.txt and independently verify that future implementation satisfies it. New requirements here remain draft, not live controls.

## DESIGN AMENDMENT 2026-11-03 — RESPONSIBILITY AND ONE WORKER CONTACT DOCUMENT
REQ-AI-40-ROLE-001: Guiding AI RFCs own WHAT/WHY, Standards own interoperable exact contracts, Interfaces own HOW, objectives own actual task and human authority remains independent. The future worker receives ONE contact document, not three: current worker Continuity is required, current Context is an optional substantive second section, and current authenticated Assignment is an optional LAST section. Optional onboarding/qualification is not a fourth required section.
REQ-AI-40-ROLE-002: Context is not a task plan; assignment is not a substitute for identity or operating environment. Every cross-RFC conflict MUST name one accountable Standard/Interface owner and an independently verifiable resolution, rather than duplicate the implementation mechanism.