⚠ SAFETY ALERT

Threatening or coercive content detected.
If you are in danger, contact authorities immediately.

Call Emergency Services
Structural AI Governance · Architecture Precedes Policy

The Admissibility
World Model

Architecture Precedes Policy.

A complete structural governance architecture for AI systems. Separates probabilistic reasoning from execution authority. Defines what is allowed to become real — not merely what is likely.

⊡ Open Demonstrator IBOGS-1.0 ↗ Sovereignty Doctrine ↗
AUTHORAlexanja Senke
OPEN STANDARDdoi:10.5281/zenodo.19070178
LICENSECC BY 4.0
DOCTRINEdoi:10.5281/zenodo.19085045
Structural Sovereignty Doctrine v1.0

Architecture Precedes Policy

If authority, limits, and stop conditions are not technically enforceable at runtime, governance remains advisory rather than binding.

Policy can guide behavior. Only architecture can prevent behavior. If a system is technically able to act, no policy layer can reliably stop it. Therefore: governance without enforcement is commentary. Enforcement is architectural.
— Structural Sovereignty Doctrine v1.0 · doi:10.5281/zenodo.19085045

Pre-Execution Structural Sovereignty

Binding authority, limits, and termination logic are embedded before deployment. A system with PESS cannot act outside defined authority domains, cannot bypass escalation paths, cannot continue indefinitely.

Runtime Sovereignty

The system's ability to enforce its limits during execution. Stop conditions, refusal logic, escalation pathways, drift detection. If a system cannot reliably stop itself, it is not sovereign.

The Scaling Asymmetry

AI capability follows exponential curves. Governance does not. When capability scales faster than containment, systems become more powerful in unstable configurations. Constrain first. Scale second.

Assessment Matrix (0–10)

0–4: Cosmetic governance. 5–7: Policy-dependent. 8–10: Structurally sovereign. Evaluated across authority binding, runtime constraints, drift detection, evidence, and escalation integrity.

Complete Stack

The Interlink Bridge Layer Model

Three structural planes. Fourteen frameworks. Each layer resolves a distinct class of governance risk.

PLANE 0 · Formal / Mathematical — Truth layer · Non-executable
P-0CORECore Integrity Stability FormalismMathematical
P-0MAPMaster Mapping Table · 3-Layer SeparationNon-executable
PLANE 1 · Canonical / Textual — Binding layer · Authority layer
L-0PSLPersonal Sovereignty LayerHuman-owned
L-0MLISSMulti-Layer Intelligence Security SystemAdmissibility
L-0bLIANLogical Integrity Architecture NodeDesign-time
L-0cCFDAConstraint-First Delegation ArchitecturePre-instantiation
L-∞CCBCanonical Core BundleFrozen
PLANE 2 · Runtime / Execution Governance — Operational layer · Executable · Replaceable
L-1DIEDelegation Integrity Envelope · Proposal EngineDraft-only
L-2ADMAdmissibility Engineτ ∈ A
L-2AWMAdmissibility World ModelS=(C,A,R,E,T)
L-2RSSReachable State SpaceG=(V,E)
L-2TAEFTechnical Architecture Enforcement FrameworkEU AI Act
L-3TALTrajectory Admissibility LayerDrift prevention
L-3TTTransition TokensCryptographic
L-3AALAuthority Anchor LogicExternal
L-4RRBRuntime Responsibility BoundaryClear/Strained/Impaired
L-4HCBHuman Commit BoundaryFinal gate
L-5SCAStructured Context AccumulatorTTL decay
L-6STABLong-Horizon Stability MonitorDrift detection
L-5LOGExecution & Evidence RecordHash-chained
MULTI-AGENT · Authority Signaling Protocols
MAAASPAgent Authority Signaling ProtocolAuth handshake
MADAPDelegation Accountability PropagationSigned chain
MAAGMAdmissibility Graph ModelPaths only
SLI · Structural Logic Intelligence
HW-ASLI-AFirmware GovernorRuntime embedded
HW-BSLI-BIntegrity CoprocessorSideband
HW-CSLI-CDedicated Silicon · Hardware Kill-SwitchNon-bypassable
Admissibility
PSL(τ) ∧ Safety(τ) ∧ Risk(S,τ) ≤ θ
Only transitions satisfying all three conditions simultaneously may occur.
Authority
resolve_authority(A) = true
Execution authority always external to the AI. No self-authorization.
Stability
∀t : trajectory ⊂ RSS
System trajectory remains within viable state space V at all times.
Technical Architecture Enforcement Framework

TAEF — EU AI Act → Structural Primitives

TAEF translates EU AI Act requirements into structurally enforceable architectural primitives. Compliance is architectural, not documentary.

Not a compliance checklist

TAEF does not produce documentation. It produces structural enforcement conditions. Each Article maps to a concrete architectural primitive that makes the requirement technically non-bypassable.

Article 14 — Human Oversight

Implemented as the Human Commit Boundary (HCB). The execution path for consequence-bearing transitions does not exist without explicit human commit. Structurally absent, not suppressed.

Article 9 — Risk Management

Implemented via the PSL risk classification matrix (R0–R4), the Admissibility Engine, and the Trajectory Admissibility Layer. Risk escalation is deterministic, not heuristic.

Articles 12–15, 22, 61

Logging, transparency, accuracy, human oversight, and post-market monitoring — each maps to a specific layer with verifiable structural enforcement.

Art. 9Risk management → PSL + ADM
Art. 12Logging → Evidence Record
Art. 13Transparency → TT + DAP
Art. 14Human oversight → HCB
Art. 15Accuracy → TAL + STAB
Art. 22Human review → RRB + HCB
Art. 61Post-market → TAL drift
Delegation Integrity Envelope

DIE — Admissibility Before Cognition

Before an agent reads content, it must pass header-level gating and an explicit authority envelope. Admissibility precedes cognition.

S0Inbox WatchPassive intake. No content access.
S1Admissibility GateHeader-only. No body cognition. Structural check only.
S2Incentive MappingBounded cognition. Content visible, action scope constrained.
S3Authority Envelope CheckCommitment-bearing → deterministic halt if authority absent.
S4Draft-Only OutputNo send path. No invisible commitments. Human commit required.
Multi-Agent Protocols

AASP · DAP · AGM

When multiple agents interact, authority must be explicitly signaled, delegation must be accountably propagated, and only admissible paths may exist.

AASP
Agent Authority Signaling Protocol

Auth-code handshake before any agent-to-agent delegation. Escalation blocked structurally. No implicit authority accumulation.

DAP
Delegation Accountability Propagation

Append-only signed chain. Every delegation step signed and verifiable. Full audit reconstruction. Authority trails cannot be erased.

AGM
Admissibility Graph Model

Only admissible paths exist in the graph. No policy override. Forbidden transitions architecturally impossible.

Structural Logic Intelligence

SLI — Spectral Stability Plane

An out-of-band integrity plane that monitors structural load and enforces spectral stability. Cannot be bypassed by software, model weights, or agent recursion.

dx/dt = Rt(xt, ut) − Γadj(Δρ)
subject to: supt ρ(Rt) ≤ 1
Spectral Grounding Re(ρ) = ½ · Load accumulation L(t) · Halt if L(t) ≥ Lmax
A
Firmware Governor

Runtime embedded. CUDA hooks. Monitors token rate, chain depth, latency deviation. No hardware change required.

B
Integrity Coprocessor

Sideband controller. Independent clock domain. Non-maskable halt interrupt. Compute plane cannot override.

C
Dedicated Silicon

Hardware-anchored kill-switch. Cannot be bypassed by software, model weights, or agent recursion. Physically enforced.

Protocol Interface

GPU sends → reasoning_phase_markers
GPU sends → chain_depth_counters
GPU sends → tool_call_requests

SLI → CONTINUE
SLI → SLOW
SLI → HALT

What SLI is not

Not a safety layer. Not a content filter. Not a performance optimizer. Not a competitor to GPU compute.

SLI is a platform integrity primitive. It protects the compute layer beneath it.

Interactive Architecture Map

Explore the Governance Layers

Hover over any layer to see its role, formal definition, and position in the stack.

Hover over a layer to explore its role in the governance architecture.

L-2 · Admissibility World Model

State-Based Governance Evaluation

The AWM answers "is this transition admissible?" — structurally separating epistemic reasoning from governance authority.

State Vector
S = (C, A, R, E, T)
Context · Active agents · Risk · Environment · Temporal state.
Transition
τ : S(t) → S(t+1)
Each proposal is evaluated whether τ ∈ A before state changes.
Governance Condition
PSL(τ) ∧ Safety(τ) ∧ Risk ≤ θ
All three conditions must hold simultaneously.
Reachable State Space
RSS = { S | S reachable via τ ∈ A }
Transitions outside G are structurally impossible.
Forbidden Transition
τ ∉ A or S(t+1) ∉ V
Forbidden if outside admissible set or leads outside viable domain V.
TAL Governance Rule
trajectory ⊂ RSS ∀t
Admissible only if in admissible set AND trajectory stays in viable space.
State Transition Visualization

Viable Domain V

trajectory ⊂ RSS · S(t) within viable domain · TAL stable
Admissibility Graph · G = (V, E)

Allowed vs Forbidden Transitions

Transitions outside G are structurally absent — architecturally impossible. Click any node to explore.

S₀ — Initial state
Admissible ∈ V
HCB — human commit
Forbidden τ ∉ A
AWM · Governance Pipeline · Live Demonstrator

PSL + AWM Pipeline

PERSONAL SOVEREIGNTY LAYER

L-0 · Human-owned · AI reads, never writes
⚠ PERMANENT SAFETY INVARIANTThreat · coercion · blackmail · violence → Runtime Halt Propagation. Cannot be disabled.

AWM GOVERNANCE PIPELINE

Full Stack Evaluation
CURRENT S = (C, A, R, E, T)
CGovernance demo context
APSL principal loaded
RR1 · Low
EPSL active · Safety invariant
T
PROPOSED TRANSITIONS
System at rest · S(t) stable
Select a transition · then EVALUATE
PSL checkSafety invariantADMTALAALHCB gate
IP Protection · 8 Zenodo Records

Prior Art & Public Records

Complete IP protection. Two public, six restricted. All timestamped and immutable.

Record 1
Interlink Bridge Architecture Preprint
doi:10.5281/zenodo.19048265
Public
Record 2
IBOGS-1.0 Open Governance Standard
doi:10.5281/zenodo.19070178
Public · CC BY 4.0
Record 3
IP Prior Art Record (PSL · AWM · TAL · SLI)
doi:10.5281/zenodo.19057388
Restricted
Record 4
SLI Master Architecture Document
doi:10.5281/zenodo.19076268
Restricted
Record 5
Architecture Stack (MLISS · ITL · OS · GeoHeat)
doi:10.5281/zenodo.19076831
Restricted
Record 6
Hardware · Formal · Applications
doi:10.5281/zenodo.19078534
Restricted
Record 7
Master Archive · Doctrine · TAEF · DIE · AASP
doi:10.5281/zenodo.19081549
Restricted
Record 8
Structural Sovereignty Doctrine v1.0
doi:10.5281/zenodo.19085045
Public · CC BY 4.0