The M.I.A.M. Doctrine: Mutual Inhibition of AI-Mediated Escalation

Public bilingual research release · v1.0

The safest system is not the fastest one, but the one whose structures force it to slow down before the button.

Broadcast 003 · Human authority and dignity · EN 52:45 · VI 22:01

Why Human Authority Cannot Be Automated

A second full-length listening companion to The M.I.A.M. Doctrine—on the human authority that must remain non-delegable when machine judgment enters an irreversible decision.

Phẩm giá con người thời AI đặt câu hỏi tương ứng từ phía con người: phẩm giá, trách nhiệm và quyền quyết định nào tuyệt đối không thể bị tự động hóa.

Why Human Authority Cannot Be Automated · 52:45

Download English M4A ↓

Phẩm giá con người thời AI · 22:01

Tải bản M4A tiếng Việt ↓

Broadcast 002 · Full-length audio · 49:03

A Nail Tech Rewrites Nuclear AI Strategy

The listening companion to The M.I.A.M. Doctrine—a spoken journey through mutual inhibition, preserved human authority and the discipline of slowing down before the nuclear button.

Bản nghe đồng hành cùng Học thuyết M.I.A.M.—từ kiềm chế lẫn nhau, thẩm quyền con người đến kỷ luật buộc hệ thống phải chậm lại trước nút bấm hạt nhân.

A Nail Tech Rewrites Nuclear AI Strategy · 49:03

Download English M4A ↓

Thợ nail phản biện AI hạt nhân · 23:16

Tải bản M4A tiếng Việt ↓
© 2026 Tony T. Nguyen | Nguyễn Từ Vĩnh TrườngCopyright © 2026 - Tony T. Nguyen | Nguyễn Từ Vĩnh Trường - Highland Reflexive Press - All rights reserved

01 · The strategic reversal

Keep the world
from becoming
its last artifact.

“Souvenir Intelligence” names an intelligence disciplined to preserve a world that can still be remembered by living witnesses—not one reduced to ash with no one left to record it.

M.I.A.M. begins where machine-speed strategy becomes most dangerous: the interval in which warning, recommendation, authorization, and irreversible force can collapse into one another.

The doctrine follows the three-part architecture of the 2025 MAIM proposal, then places a counter-principle against each pillar. Its concern is not only who may dominate advanced intelligence. It is whether an attempt to prevent domination—or a false alarm about it—can become the last day of the world.

01

Mutual inhibition, not mutual sabotage.

The default objective is zero launch: no weapon fired, including in retaliation, until human authority has survived the warning process.

02

No delegation of irreversible authority.

The protected object is not only the weapon or model. It is the decision itself, which must not pass silently from human hands into a machine’s.

03

Deliberate slowness, not the race for speed.

Safety is produced by architecture: pauses, independent corroboration, and authority checks that remain effective under pressure.

02 · MAIM / M.I.A.M.

Two zeros.
Different worlds.

MAIM’s zero is strategic paralysis: no actor can pull decisively ahead. M.I.A.M.’s zero is operational non-occurrence: the button is not pressed.

MAIM · Mutual Assured AI Malfunction

Deterrence through vulnerability.

Primary object
Prevent unilateral superintelligence dominance.
Control logic
Mutual sabotage and credible counter-capability.
Temporal pressure
Compete, verify, and respond before rivals pull ahead.
Failure feared
One actor escapes the balance and gains decisive advantage.
M.I.A.M. · Mutual Inhibition

Restraint through preserved authority.

Primary object
Prevent AI-mediated escalation from becoming irreversible force.
Control logic
Pause, corroborate, inhibit, and keep authorization human.
Temporal pressure
Build slowness into the path from warning to consequence.
Failure feared
A false signal, delegated response, or race dynamic ends the world.

03 · Objections in the record

Restraint must
survive its critics.

The doctrine treats objections as part of the architecture, not as decoration added after the conclusion.

01

Does slowness weaken deterrence?

The paper distinguishes delay from paralysis: designed friction can preserve credible authority while preventing an automated rush from warning to launch.

02

Can humans remain in the loop at machine speed?

M.I.A.M. rejects a ceremonial human role. Authority must be operationally current, independently informed, and able to stop the sequence before irreversibility.

03

Can mutual inhibition be verified?

The burden shifts from trusting intentions to inspecting procedures: corroboration paths, pause conditions, delegation boundaries, and accountable release authority.

04 · Complete bilingual web edition

The M.I.A.M. Doctrine Học thuyết M.I.A.M.

The M.I.A.M. Doctrine © 2026 Tony T. Nguyen | Nguyễn Từ Vĩnh Trường · Highland Reflexive Press · All rights reserved

Đang mở toàn văn tiếng Việt…

Opening the complete English text…

05 · Bilingual reader editions

One doctrine.
Two complete routes in.

Both reader editions are public under one canonical research record and one DOI.

Bilingual English and Vietnamese M.I.A.M. reader edition cover
HRP–RN–013–EN · 58 pages

The M.I.A.M. Doctrine

Mutual Inhibition of AI-Mediated Escalation — A Souvenir Intelligence Strategy.

Open English PDF
Bìa ấn bản tiếng Việt M.I.A.M.
HRP–RN–013–VI · 46 trang

Học thuyết M.I.A.M.

Kiềm chế lẫn nhau trước leo thang qua trung gian trí tuệ nhân tạo — Chiến lược Trí tuệ Kỷ vật.

Mở bản PDF tiếng Việt

06 · Canonical record

Cite the release.
Keep the version.

This page and both updated PDFs identify the work as HRP–RN–013, version 1.0. The DOI continues to identify the canonical research record.

Author
Tony T. Nguyen | Nguyễn Từ Vĩnh Trường
Publisher
Highland Reflexive Press
Record
HRP–RN–013
Version
1.0 · November 14, 2026
Form
Bilingual public research report
Status
Independent · not peer reviewed
Digital Object Identifier10.5281/zenodo.24215109 ↗
Suggested citation

Nguyen, Tony T. (Nguyễn Từ Vĩnh Trường). (2026). The M.I.A.M. Doctrine: Mutual Inhibition of AI-Mediated Escalation—A Souvenir Intelligence Strategy (Version 1.0). Highland Reflexive Press. https://doi.org/10.5281/zenodo.24215109

Authorship & AI assistance. Tony T. Nguyen originated the doctrine, governing distinctions, argument, and final publication decisions. Generative AI materially assisted comparison, analytical development, drafting, editorial preparation, and production under the author’s direction. Responsibility for the claims and final text remains with the human author.