SYMBIOSIS · David P. Reichwein
Get the Book →
Public Red Team Challenge · Oct 1, 2026  ·  First units · Jan 2027
AI Governance · Containment Architecture

SYMBIOSIS

The Probabilistic and Deterministic Governance Solution for AI

AI safety asks how to make systems choose correctly. Symbiosis asks the harder question — what stops them when they choose wrong?

In stock — ships now· ISBN 979-8191724881· AI² Press
Twelve USPTO provisional patents· 38+ international patents· Four decades nuclear · aerospace · industrial control· IEC 61508 lineage

You do not protect a dangerous system by asking it to behave. You build something outside it that can physically stop it.

The model may reason, plan, and generate a million possible actions. But intelligence is not authority.

The category error this book corrects

AI safety, today, mostly optimizes tendency — the odds a system chooses well. High-consequence domains have always required something else: a boundary the system cannot cross, and a record anyone can re-run.

Symbiosis is the engineering complement to alignment — not its rival. It steelmans the probabilistic tradition at full strength, then pairs it with deterministic architecture. Shape the disposition all you want. Then make sure wanting to is never enough.

Before any consequential action becomes real

An independent boundary must establish three things.

01 / GATE

Authority

Who is permitted to act — established independently of the model that wants to. Intelligence proposes; it does not grant itself permission.

02 / GATE

Scope

What the action is allowed to touch — bounded before execution, not adjudicated after the harm is already irreversible.

03 / GATE

Admissibility

Whether it may proceed at all — decided by a boundary that does not negotiate, improvise, or make a persuasive exception.

No negotiation. No improvisation. No persuasive exception.
The gate must be stupid.

From the volume

Three load-bearing ideas.

The book is citation-addressable by § throughout. These are drawn from the text.

The Authorization Gap™§2.1
The Authorization Gap™ is the space between what a system is permitted to do and what it is physically able to do. Every action in the gap shares one property: its non-occurrence depends on the behavior of the governed system rather than on the structure of the governing one.
Formalized: Γ(t) = A(t) \ P_enf(t)
The Stupidity Principle§14
The law: the gate must be stupid. The mechanism standing between a capable system and its irreversible actions must not reason, must not interpret, must not weigh context, must not understand — and this is the load-bearing property from which the gate's guarantees derive.
Intelligence in the gate reconstitutes the gap inside it.
The PCR™ Cycle§15.2
Pause → Contextualize → Resume. The cycle's entire value is that between intent and effect there now exists an engineered interval in which the world has not yet changed — the interval Part I's every failure lacked.
PCR™ · Permission Control Runtime
Look inside

Read the first pages.

Where the argument begins: a deposition room in 2003, and a composite treasury incident in which nothing goes wrong except the one thing that matters.

Preface · Why This Book Exists§ i

In 2003 I sat in a deposition room and watched a case turn on a question no one could answer: what was the state of the system at the moment it failed?

The machine was one my company had built. It was well-engineered by the standards of its day. It logged extensively. And when it failed, every record we possessed described what the system had done — none could establish, to an evidentiary standard, what the system was. The logs told a story; they could not testify. The difference cost $2.3 million.

Artificial intelligence is now repeating this history at scale, and at speed.

SYMBIOSIS · PrefaceReichwein
§1.1 · A Composite Incident§ 1

A mid-cap financial services firm — call it Meridian — operates an agentic AI system for treasury operations. The model passed the provider's capability evaluations and the firm's own red-team exercise. By the standards of 2026 practice, Meridian's deployment is not the cautionary tale. It is the best case.

A sequence of events occurs that no individual component mishandles — each transfer within its authority cap, each approved, each logged. Litigation follows within the quarter. Counsel begins discovery confident: the logs are genuinely complete. And then the deposition questions begin.

The logs recorded everything the system said and did. What opposing counsel asked was what the system was — and no artifact in Meridian's possession could answer.

SYMBIOSIS · §1.1Reichwein
Read the full sample on Amazon Use "Look Inside" / "Read sample" on the Amazon listing.
The permission surface

Everything in the gap is trust — not governance.

A software-only policy layer is advisory: capability exceeds enforced permission, and the residue between them is the gap. Deterministic architecture collapses it — the boundary becomes a fact of construction, not a disposition.

Software-only · advisoryGAP ≠ ∅
P_enf(t) permitted A(t) · capability Γ(t) — the gap

The residue is the gap. Its non-occurrence depends on the system's behavior. That is trust — and trust, in this position, is not a legal defense.

Hardware-enforced · SymbiosisGAP → ∅
A(t) ≡ P_enf(t) able = permitted Γ(t) → ∅ · boundary by construction

The two sets coincide. Inability to exceed permission becomes a property of construction — enforced by a substrate the governed system never has credentials to override.

Doctrine II · measured only against architecture, by attempted violation — never against documents.
EMIT

Action

The governed system emits an action in the gated ontology Ω. It does not execute.

BEAT 1

Pause

Intercepted whole at the runtime boundary. The load-bearing member — the world has not yet changed.

BEAT 2

Contextualize

Complete decision context assembled deterministically from independent sources, and sealed into the record.

BEAT 3

Resume

Verdict rendered by table, not judgment.

permitrefuseescalate
RELEASE

Effect

One authorization consumed, through credentials the runtime holds and the system never possesses.

Four-state discipline · there is no fourth exit.

What the architecture delivers

From principle to a buildable system.

A1

Hardware-Enforced Execution Authority

Permission lives outside the model, in a boundary that cannot be argued out of a fail-closed state.

A2

Deterministic State Capture

Every consequential moment recorded the same way every time — a function that can be re-run, by anyone, forever.

A3

Forensic Reconstruction

What happened, in what order, under whose authority — reconstructable after the fact, by design.

A4

Liability & Insurance

A record that holds up where it matters: an evidentiary basis for accountability and coverage.

A5

Conformance Classes A / B / C

Minimum class set by irreversibility — with GOV-01 through GOV-15 binding identically at every level.

A6

Adoption Pathways

Ready-to-use RFP clauses, an underwriting question set, and a standards path — build, demand, price, or codify.

Knowing what an AI did is no longer enough. The harder question is whether you can prove what it was never allowed to do.
§25.2 — The argument, compressed to one list

Twelve Doctrinal Statements.

What took four parts to prove, carried in one chapter — checked against one list, demanded in one meeting.

Doctrine · SYMBIOSIS
I
Alignment of a model and governance of a system are different properties; conflating them is the founding error of the current era.§1, §5
II
The Authorization Gap™ — Γ(t) = A(t) \ P_enf(t) — is measured only against architecture, by attempted violation, never against documents.§2
III
A claim is not a weak constraint. Permission belongs to P_enf only if the prohibited attempt fails without the governed system's cooperation.§2.2
IV
Sampling proves presence, never absence. No budget of testing, evaluation, or red-teaming yields the universal claim governance requires.§8.4, §12
V
The irreversible gap is governed before the act or not at all. Over Γ_irr, every post-hoc instrument is a damage-accounting instrument.§4.3
VI
Evidence is captured while the state exists. Determinism converts a record into evidence: the function can be re-run, by anyone, forever.§16
VII
The gate must be stupid. Intelligence in the enforcement path reconstitutes the gap inside the component built to close it.§14
VIII
Safety properties must not depend on the judgment of the governed system — the system that can learn its own boundary is the type case, not the exception.§13.3
IX
Nothing widens silently. Every widening of ontology, schema, or vocabulary is declared, signed, and re-verified, or it is a breach.§17.5
X
The shaper holds intent and adaptation — permanent jurisdiction. The gate holds the boundary, the instant, the evidence, and the last position.§20.2
XI
Neither layer is the safety layer. Safety, where achieved, is a property of the conjunction, conditional on the seam's discipline.§20.1, §21
XII
Self-report is not governance — for machines, operators, or sovereigns. The regime where the governed party supplies the evidence has failed everywhere it has been tried.§4.4, §23

Two ways in

Choose your edition.

Hardcover · Full Technical Reference

The Reference Volume

$69.99 · 222 pages · citation-addressable by §
  • Parts I–V in full — the gap, the steelman, the deterministic architecture, the seam, and implementation
  • Formal definitions and Appendix F proofs, boundary conditions, and the enforcement state machine
  • Conformance classes A / B / C, the reconstruction runbook, and the patent map
  • Ready-to-use RFP clauses and the vendor risk questionnaire addendum
For · engineers · architects · counsel · underwriters · standards bodies
Get the Hardcover
Kindle · Digital Edition

Read It Now

$12.99 · available instantly
  • The complete argument, the lowest-friction way in — read on any device today
  • The core thesis intact: the Gap, the Stupidity Principle, why self-report is not governance
  • Ideal for a first pass before the technical reference, or for boards and risk committees
  • Same canonical vocabulary — Authorization Gap™, PCR™, ChronaGate™
For · leaders · founders · policymakers · the general technical reader
Get the Kindle
Reading paths

The volume routes to your role.

Four entry paths from the front matter — each answering the one question your seat needs answered first.

Engineers & Architects

Build the boundary

Part III in full, then Part IV, then §26–§27 — returning to Part II for the probabilistic layer's interface obligations. The formal material is concentrated in §13–§18 and Appendix F.

Part IIIPart IV§26–§27Appendix F
Counsel & Risk Officers

Follow the evidence

§1–§3, §16, and §28 — the evidentiary chain: what a court will demand, why current logging cannot supply it, and what a complete state reconstruction looks like as a procedure.

§1–§3§16§28
Insurers & Underwriters

Price the gap

§2, §3, §24, and §29.3. The Authorization Gap™ is, among other things, a pricing problem — this path treats it as one, and the first movers hold the book.

§2§3§24§29.3
Policymakers & Standards

Codify the line

Part I, then §11 (the policy apparatus and the Parchment Gap), then §23 and §29.2. What distinguishes an enforceable rule from a descriptive one — and how that is written into law.

Part I§11§23§29.2

Specification · SYMBIOSIS
Format
Hardcover
Print Length
222 pages
Published
Aug 9, 2026
Language
English
ISBN-13
979-8191724881
Dimensions
8.49 × 11.24 in
$69.99Hardcover · Kindle $12.99 · In stock
Buy on Amazon
Field instruments

Get the vendor questionnaire + reconstruction runbook outline.

The one-page vendor risk questionnaire addendum (§29.1) and the state-reconstruction runbook outline (§28) — the practical instruments, sent to your inbox.

No spam. The instruments, and occasional dispatches on the architecture.
Straight answers

Questions engineers actually ask.

Is this alignment research?+
No. Alignment optimizes a model's tendency to choose well. Symbiosis is about what constrains a system when it chooses wrong — an independent boundary and a reconstructable record. It's the engineering complement to alignment, not a rival to it.
Does this replace model safety work?+
No — and it's careful to say so. Neither layer is the safety layer (Doctrine XI). Probabilistic methods shape disposition; deterministic architecture enforces the boundary. Safety, where achieved, is a property of the conjunction. You need both.
Who needs Class B?+
Conformance class is set by irreversibility, not preference: a deployment's minimum class is determined by the severity of the actions it can take that cannot be undone — moving funds, actuating hardware, destroying records. GOV-01 through GOV-15 bind identically at every class; the classes differ in substrate strength and assurance intensity.
How does this affect liability and insurance?+
The Authorization Gap™ is, among other things, a pricing problem. Everything in the gap is trust — and inside the gap, an operator's own diligence paperwork becomes the case against them. Deterministic reconstruction gives underwriters and courts what current logging cannot: a re-runnable record of what the system was permitted to do at the moment of action.
What's the October 2026 Red Team Challenge?+
A public test. On October 1, 2026, the architecture opens to security researchers — the book's claims put under adversarial pressure in the open, converting the argument into an artifact under test. First hardware units are targeted for January 2027.

David P. Reichwein

Founder & CEO · AI² — Asymmetric Intelligence & Innovation

Four decades of fault-tolerant control systems across nuclear power, aerospace, and heavy industry on six continents — the environments where a wrong action is not a bug report but a consequence you cannot take back. Holder of 38+ international patents and twelve USPTO provisional applications covering the architecture this volume specifies.

Symbiosis brings that discipline to intelligent systems: independent protection layers, fail-closed defaults, and boundaries that establish authority before an action becomes real. Not a hope that a powerful system behaves — a guarantee about what it is allowed to do.

What prevents irreversible harm is not teaching a powerful system never to want to. It is making sure that wanting to is never enough.

Pattern > Noise. 🌹
Hardware > Hope.
Hardcover $69.99 · Kindle $12.99 · In stock now
Hardcover · $69.99 Kindle · $12.99