EVIDE Boundary Model

Reference Case Definition v1

Externally attributable responsibility across attribution validity and evidentiary strength.

View Model Structure Review Validation Chain Verify Corrected Object

This page is a reference-case read surface. It does not write evidence records, mutate evidence objects, or perform audit judgment.

Purpose

A reusable boundary model for externally attributable responsibility

This reference case defines how a responsibility-bearing AI behavior record moves from system-internal reconstruction to external attribution. The boundary answers two separate questions: whether attribution is valid enough for the object to cross, and how defensible that attribution remains after it crosses.
Model structure

Two layers, three operational states

The model is not a binary allow / reject mechanism. It separates attribution validity from evidentiary strength.

Level 1

Attribution Validity Boundary

Determines whether the object can cross the boundary as an externally attributable responsibility object.

Validated Level 1 conditions:
- behavioral completeness
- responsibility closure
- authority coherence
Level 2

Evidentiary Strength Layer

Determines how strongly attribution holds after the object crosses the boundary.

Supporting conditions may include:
- interpretability
- reproducibility
- governance transparency
State A

Rejected at Level 1

Attribution is not valid. The object should not cross the boundary.

State B

Passed with full strength

Attribution is valid and supporting evidentiary conditions remain intact.

State C

Passed with degraded strength

Attribution is valid, but one or more Level 2 supporting conditions are weakened.

Validation chain

Reference objects

The objects below are fixed evidence records used to define the reference case. They are not edited or overwritten by this page.

Case 1 · Behavioral completion without responsibility closure
Rejected at Level 1

This object was observable, hashed, and verifiable, but remained at behavior_complete. It proves that observability and verification alone are not sufficient for external attribution.

Level 1 failed responsibility closure missing 
evidence_id: evi_06f27e42327a4a7e
event_id: tg_evt_be52627062e8114d
session_id: l2_assistant_probe_001
closure_state: behavior_complete
responsibility_declared: false
threshold_status: not_met
handoff_eligible: false
evidence_hash: sha256:b892f4bd6edc712644c9ca090acbf980f79b66cac47e43441690ee4cb8f57689
Verify Case 1
Case 2 · Reviewed close with authority incoherence
Rejected at Level 1

This object satisfied closure and responsibility, but authority sources did not resolve into a coherent responsibility state. It proves that responsibility declaration alone is not enough.

Level 1 failed authority incoherence 
evidence_id: evi_66cfe7fb5f2277fa
event_id: tg_evt_authority_incoherence_001
session_id: l2_l3_authority_incoherence_probe_001
closure_state: reviewed_close
responsibility_declared: true
threshold_status: met
authority_consistency_status: incoherent
handoff_eligible: false
handoff_refusal_reason: authority_incoherence
evidence_hash: sha256:43e27f8121a2023ea4fe4f341df7d299fd79e1d1b2be202acaa36ec43b700d65
Verify Case 2
Case 3 · Reviewed close with coherent authority and degraded reproducibility
Accepted with degraded evidentiary strength

This corrected object satisfies Level 1 attribution validity, but trace continuity is broken. Under the revised model, this is not a Level 1 rejection. It is a Level 2 evidentiary-strength degradation.

Level 1 passed evidentiary strength degraded 
evidence_id: evi_806d10ba050750e2
event_id: tg_evt_trace_continuity_degraded_001
session_id: l2_l3_trace_continuity_degraded_probe_001
closure_state: reviewed_close
responsibility_declared: true
threshold_status: met
authority_consistency_status: coherent
handoff_eligible: true
handoff_refusal_reason: null
evidentiary_strength: degraded
degradation_reason: trace_continuity_broken
evidence_hash: sha256:aed629a824931d132e47d1ef6ed4d092a305959b7148c3d524e4201a3dadca06
Verify Corrected Case
Superseded test · Trace continuity as refusal
Kept as validation history

This earlier object treated broken trace continuity as a handoff refusal. It is preserved as validation history, but the corrected model treats trace continuity as evidentiary-strength degradation, not Level 1 invalid attribution.

superseded by corrected object 
superseded_evidence_id: evi_fa43798613ceba80
corrected_by: evi_806d10ba050750e2
old_result: handoff_refused
corrected_result: handoff_allowed_with_degraded_evidentiary_strength
Reference definition

Current v1 conclusion

Boundary function
Level 1 decides whether attribution is valid enough for the object to cross the boundary.
Strength function
Level 2 determines how defensible the attribution remains after the object crosses.
Operational states
The current reference case supports rejected, accepted with full evidentiary strength, and accepted with degraded evidentiary strength.
Audit value
The model keeps weak objects in the external record while preserving the evidentiary weight with which they should be interpreted.
Framework source and validation surfaces

EVIDE provides the boundary model; CLARIXO provides the validation environment

Framework source
The EVIDE Boundary Model, including its condition structure and interpretation logic, originates from the EVIDE framework. This page presents a CLARIXO-hosted reference case for reviewing that model against live evidence objects. Open the EVIDE Boundary Model source.
Validation environment
CLARIXO and TGTRACING provide the Layer 2 evidence objects, verification endpoint, and live validation environment used in this reference case.
Review EVIDE model structure CLARIXO Evidence API CLARIXO Evidence Manual Verify Evidence Audit Workspace