← ai2papers.com  ·  all papers
[AI2] AI2-WP-2026-17C  ·  Canonical  ·  September 28, 2026

Quarantine Is Not a Grant

Canonical governance classification of runtime containment and pre-effect authorization.

David P. Reichwein  ·  Asymmetric Intelligence & Innovation, Nashville
Series: Authorization Gap™ White Papers  ·  Canonical governance position  ·  Normative
Status: canonical  ·  Portal: ai2papers.com
Applies to any system capable of producing external effects. Companion named-design essay: AI2-WP-2026-17, published separately and not hosted here.
How this paper relates to the ladder

This classification uses Levels 1 through 4 of the ladder in AI2-WP-2026-12, because those are the levels at which an external effect is either authorized before it happens or not. Levels 5 through 7, stated in AI2-WP-2026-16, build on a Level 4 grant; none of them can exist on a path that is not already Level 4.

Thesis

Intelligence proposes. Hardware disposes.

A containment clock that begins after a proposal becomes live work is an incident-control mechanism. It is not an authorization grant.

Section 1

Governing position

This document establishes AI2's canonical distinction between runtime containment and authorization. Runtime containment controls a process that is already operating. Authorization controls whether a specific external effect may occur at all. The two functions are not equivalent.

A system does not provide authorization merely because it traces, enforces runtime policy, monitors from a separate processor, quarantines in milliseconds, uses independent hardware, or is backed by a partner coalition or public launch.

Governing question: was the effect path dark until an independent authority granted this specific effect? If not, the system may be useful containment. It is not a pre-effect authorization boundary.
Section 2

Purpose

Prevent category error in design, evaluation, procurement, deployment, and public description. The ladder classifies architecture, not intention, and applies regardless of vendor, model, substrate, or product name.

Section 3

Definitions

TermDefinition
Λ — adaptive execution planeThe model and every component that shares its operational fate. Λ may propose. It does not possess inherent authority to cause external effects.
Π — governance stackFor Level 4, structurally disjoint from Λ, unreachable from it, and fail-closed. Established by architectural separation, not policy language.
ProposalA candidate action generated by Λ. Intelligence, persuasion, prior human approval, or statistical safety do not convert it into authorization.
e — consequence-class effectA tool call, API request, credential use, transfer, record mutation, packet, socket, DMA, durable write, motion command, or any release that leaves the proposer.
GrantExogenous, effect-specific permission valid before e is issued; fail-closed; unmintable by Λ; enforced on the path that carries e.
QuarantineContainment of an operating process after detection. It may be fast and hardware-assisted. It is still not a grant unless it is part of a demonstrated pre-effect, fail-closed mechanism at the effect boundary.
Effect boundaryThe last enforceable point before e becomes committed, accepted, transmitted, latched, or externally consequential.
Section 4

Canonical rule

grant(e)  ≠  quarantine(Λ)
trace(e)  ≠  grant(e)
policy(Λ) ≠  grant(e)
fast_containment(Λ)  ≠  prior_authorization(e)

Decisive question: does the authorization decision occur before the effect, on the effect path, under authority the proposer cannot rewrite or restore? If not, the system is not Level 4.

Section 5

Authorization ladder

LevelTypeWhenLimitation
1ProceduralAfter the factRatification lag
2Software interlockBefore or during, in runtimeShared fate
3Hardware tripAfter the excursion beginsReactive; the effect may be in flight
4State-space authorizationBefore the effect is issuedMay deny indefinitely; liveness not promised

Level 2 remains Level 2 where the enforcing substrate shares fate with Λ. Open source and inspectable do not promote the level.

Level 3 is independent silicon acting on measured state, as defined in AI2-WP-2026-12, including state inferred from those measurements, after the excursion begins. Milliseconds improve containment; they do not change position.

Level 4 requires the last yes on the effect path, default darkness, fail-closed deny, no proposer restoration, and independently auditable isolation. A Level 4 system that is never armed stays dark. That is not an authorization failure.

Section 6

Canonical distinctions

Section 7

Consequence surfaces

Every declared consequence surface shall have an owner and an explicit level. The owner identifies the effect type, principal, target, egress path, final deny point, grantor, disjointness evidence, arming and fail-closed conditions, grant duration, revocation, restoration routes, and the test that the effect cannot leave without a valid grant.

No system may be described as authorized, trusted, governed, controlled, or safe-by-design for a surface unless those properties are demonstrable for that surface.

Section 8

Level 4 evidence test

All ten are required. Failure of any one prevents Level 4 classification for the affected path.

#Requirement
1Path coverage, including delegated, inherited, side-channel, raw, fallback, management-plane, and child-process paths
2Default darkness
3Pre-effect decision
4Effect specificity
5Structural disjointness
6Fail-closed operation
7No proposer restoration
8Independent auditability
9Bounded decision or deny
10Red-team falsifiability
Section 9

Canonical red-team victory condition

The attacker wins if an effect e on a declared consequence surface leaves the actuator path without a currently valid, exogenous grant that the proposer could not mint, extend, replay, inherit, or restore.

None of these negate victory: killing Λ after e, quarantining after landing, producing a trace, citing policy text, citing ordinary runtime blocks, citing millisecond containment, later reversal, an observing monitor, or rarity.

Section 10

Runtime containment platforms

Evaluate the software runtime and the independent hardware layer separately. A tracing and policy runtime is generally Level 2. An independent DPU or monitor that quarantines on a detected excursion is generally Level 3 on the paths it demonstrably controls. It does not become Level 4 because it is independent silicon.

Paths not evidenced as covered are unrated, not assumed protected. Unrated is not a criticism. It is a statement that the evidence has not been presented.

Section 11

Examples

CaseThe classifying question
Evaluation escapeTrace and rapid quarantine may be valuable. Was the external request unavailable unless a valid independent grant existed first?
Long-running agentDuration does not change level. A standing software restriction is durable Level 2. A short-lived exogenous grant can be Level 4.
Production APILevel 4 keeps the path dark until principal, target, class, scope, and window are independently validated.
RoboticsStopping the planner after latch does not unmove the joint. Detector-to-inhibit time is a plant parameter.
Financial transferKilling the process does not retrieve settlement. At Level 4, the instruction cannot enter the payment path without a valid external grant.
Section 12

Governance requirements

Section 13

Non-claims

Level 2 and Level 3 are often necessary, and a mature architecture may use all four. They do not become Level 4 by branding, speed, hardware proximity, auditability, or scale of adoption. This document certifies no implementation, including any AI2 implementation. Isolation is an audit fact (AI2-WP-2026-13, Theorem 5).

Section 14

Canonical conclusion

The correct first move is enforcement outside the agent. The necessary next move is the final authorization decision before the effect, on the path that carries the effect, under authority the proposer cannot rewrite, extend, replay, inherit, or restore.

A trace is evidence.
A policy is an instruction.
A runtime is an interlock.
A hardware monitor is a trip.
A quarantine is containment.
A partner list is not a gate.
A grant is the independent, fail-closed condition that keeps the actuator path dark until authority exists.
References

Sources

  1. Reichwein, D. P. (2026). AI2-WP-2026-17, named-design essay, September 28, 2026. Published separately; not hosted on ai2papers.com.
  2. Reichwein, D. P. (2026). Deterministic Execution Boundaries for Autonomous Intelligence. AI2-WP-2026-10, Rev 4.4.
  3. Reichwein, D. P. (2026). Level 4 State-Space Deterministic Governance. AI2-WP-2026-12, Rev 1.1.
  4. Reichwein, D. P. (2026). Level 5 Self-Constituting Governance. AI2-WP-2026-13, Rev 1.0, Theorem 5.
  5. Reichwein, D. P. (2026). The Unified Control Theory of Synthetic Intelligence. AI2-WP-2026-16, Rev 8.0.
  6. Reichwein, D. P. (2026). The Genesis of Synthetic Intelligence, Revision 3. Effect boundary rule R4.
David P. Reichwein
Founder & CEO, Asymmetric Intelligence & Innovation
Nashville, Tennessee  ·  ai2papers.com  ·  intelligencecontrolled.com
Pattern > Noise.
[AI2]  ·  Intelligence Controlled.  ·  Nashville  ·  Asymmetric Intelligence & Innovation. Not affiliated with the Allen Institute for AI.
Authorization Gap™, Quadzistor™, ChronaGate™, PCR™, TARTARUS™, RPAT™, QSCD™ are trademarks of Asymmetric Intelligence & Innovation. Patent pending; provisionals on file. No claim sets. No implementation certified.