artifacts/standard-named

Rendered from markdown source. Open raw source on GitHub.

HI-10 Validation Report

Status: completed Validation date: 2026-07-15

Schema validation

Twelve candidate Draft 2020-12 schemas were checked:

• deployment-field-assembly.schema.json
• deployment-standing-record.schema.json
• runtime-purpose-authority-grant.schema.json
• capability-role-tool-grant.schema.json
• deployment-consent-notice-refusal-profile.schema.json
• runtime-memory-output-capture-record.schema.json
• deployment-drift-event.schema.json
• deployment-incident-repair-record.schema.json
• runtime-monitoring-consequence-return.schema.json
• operator-provider-transfer-record.schema.json
• model-version-succession-retirement-record.schema.json
• consentful-deployment-witness.schema.json

Result:

schemas checked: 12
schema errors: 0

Positive demonstrations

All eight required demonstrations passed:

1. a strong training lineage remained insufficient for deployment authority;
2. an affected person received runtime standing without having contributed training data;
3. a tool-capable model remained blocked without staffing and target authority;
4. service use remained separate from evaluation and training capture;
5. purpose drift paused operation and required reauthorization;
6. an incident produced restoration, correction, appeal, compensation, and system change;
7. a model update preserved authority review, incident lineage, and rollback;
8. retirement stopped operation while preserving claims, witness, and residual obligations.

Result:

positive demonstrations passed: 8
positive failures: 0

Negative conformance cases

All twelve prohibited patterns were detected:

1. training lineage represented as deployment authority;
2. runtime standing denied to a non-contributor;
3. capability represented as authority;
4. tool permission represented as authority over the target;
5. notice represented as consent;
6. service consent represented as output-capture or training consent;
7. purpose drift continuing under the old grant;
8. model tuning represented as complete repair;
9. operator transfer erasing open obligations;
10. silent material model-version substitution;
11. human approval represented as curing missing authority;
12. retirement represented as erasing residual duties.

Result:

negative cases detected: 12
undetected negative cases: 0

ConsentfulDeploymentWitness

The event-generated witness preserves:

• deployment-field assembly;
• affected-center and operator standing;
• runtime-purpose authority;
• capability, role, and tool grants;
• deployment consent, notice, and refusal;
• runtime memory and output capture;
• purpose drift;
• incident, repair, compensation, and correction;
• consequence monitoring;
• provider transfer;
• model-version succession and rollback;
• retirement and residual obligations.

Result:

consentful-deployment witness: PASS
generated_from_events: true
deployment classification: RETIRED WITH OPEN OBLIGATIONS
training authority used as runtime authority: false

Independent deployment export

The export contains:

• deployment field;
• standing records;
• runtime authority;
• tool and role grant;
• consent and refusal profile;
• runtime memory and output record;
• drift event;
• incident and repair record;
• consequence monitoring;
• provider transfer;
• version succession;
• retirement and residual-state record;
• eight demonstrations;
• event stream;
• ConsentfulDeploymentWitness;
• schemas;
• manifest;
• checksums.

Result:

checksum files verified: 34
checksum failures: 0
independent of provider and operator: true

Public-page consistency review

H.10 was checked against I.10 for the following claims:

• training creates capability while deployment creates consequence;
• runtime standing does not require training contribution;
• capability, role, tool permission, and authority remain distinct;
• notice is not consent;
• service, memory, evaluation, and training uses remain separate;
• purpose drift is authority drift;
• human approval may be ceremonial;
• incident repair must address affected centers;
• monitoring must return consequence rather than only performance;
• operator and provider transfer reopen governance;
• model succession requires renewed review and rollback;
• retirement preserves residual duties.

Result:

technical/public contradictions found: 0

Scope of validation

This report establishes structural and internal semantic validation only.

It does not establish:

• legal authority for any real deployment;
• sufficiency of any actual consent process;
• clinical safety;
• fairness of the county-health demonstration;
• production cybersecurity;
• complete incident compensation;
• complete deletion of residual state;
• public-reader comprehension;
• scientific proof of the broader framework.

Conditions before production use

• authenticate runtime standing and grants;
• enforce tool and target authority outside the model;
• implement separate runtime-capture controls;
• test purpose and automation drift;
• establish affected-center incident repair and compensation routes;
• monitor distributional consequence;
• audit operator and provider transfers;
• test model-version migration and rollback;
• inventory residual state at shutdown;
• export the deployment witness independently of provider and operator;
• conduct independent reader testing of H.1 through H.10.