{
  "id": "https://doi.org/10.5281/zenodo.21962425",
  "doi": "10.5281/ZENODO.21962425",
  "url": "https://zenodo.org/doi/10.5281/zenodo.21962425",
  "types": {
    "ris": "RRPT",
    "bibtex": "misc",
    "citeproc": "report",
    "schemaOrg": "Report",
    "resourceType": "",
    "resourceTypeGeneral": "Report"
  },
  "creators": [
    {
      "name": "Jacobs, Tim",
      "nameType": "Personal",
      "givenName": "Tim",
      "familyName": "Jacobs",
      "affiliation": [
        {
          "name": "KTS Global"
        }
      ],
      "nameIdentifiers": [
        {
          "nameIdentifier": "0009-0008-7130-1448",
          "nameIdentifierScheme": "ORCID"
        }
      ]
    }
  ],
  "titles": [
    {
      "title": "Evidence-Governed Intelligence: Applying Geometry Intelligence to Institutional AI Assurance"
    },
    {
      "title": "Applying Geometry Intelligence to Institutional AI Assurance",
      "titleType": "Subtitle"
    }
  ],
  "publisher": {
    "name": "KTS Global"
  },
  "container": {},
  "subjects": [
    {
      "subject": "Evidence-Governed Intelligence"
    },
    {
      "subject": "Geometry Intelligence"
    },
    {
      "subject": "Institutional AI assurance"
    },
    {
      "subject": "AI governance"
    },
    {
      "subject": "AI risk management"
    },
    {
      "subject": "Evidence governance"
    },
    {
      "subject": "Decision provenance"
    },
    {
      "subject": "Claim-level governance"
    },
    {
      "subject": "Entity identity"
    },
    {
      "subject": "Authority architecture"
    },
    {
      "subject": "Human oversight"
    },
    {
      "subject": "Human decision authority"
    },
    {
      "subject": "Constraint preservation"
    },
    {
      "subject": "Context governance"
    },
    {
      "subject": "Contradiction management"
    },
    {
      "subject": "Provenance continuity"
    },
    {
      "subject": "AI output verification"
    },
    {
      "subject": "Decision traceability"
    },
    {
      "subject": "Contestability"
    },
    {
      "subject": "Material change"
    },
    {
      "subject": "Sovereign AI"
    },
    {
      "subject": "Sovereign deployment"
    },
    {
      "subject": "High-consequence AI"
    },
    {
      "subject": "Institutional knowledge"
    },
    {
      "subject": "Auditability"
    },
    {
      "subject": "TEVV"
    },
    {
      "subject": "Generative AI assurance"
    },
    {
      "subject": "Critical infrastructure"
    },
    {
      "subject": "Regulatory evidence"
    },
    {
      "subject": "Formal verification"
    }
  ],
  "dates": [
    {
      "date": "2026-08-16",
      "dateType": "Issued"
    }
  ],
  "publicationYear": 2026,
  "language": "en",
  "identifiers": [
    {
      "identifier": "oai:zenodo.org:21962425",
      "identifierType": "oai"
    }
  ],
  "version": "Final Manuscript 1.0",
  "rightsList": [
    {
      "rights": "Creative Commons Attribution 4.0 International",
      "rightsUri": "https://creativecommons.org/licenses/by/4.0/legalcode",
      "schemeUri": "https://spdx.org/licenses/",
      "rightsIdentifier": "cc-by-4.0",
      "rightsIdentifierScheme": "SPDX"
    }
  ],
  "descriptions": [
    {
      "description": "Evidence-Governed Intelligence defines an institutional operating model for connecting machine outputs to governed entities, claims, evidence, context, authority, constraints, validation and accountable decisions.\n\nHigh-consequence institutions do not principally lack generated information. They lack a durable structure for determining what machine-generated information concerns, what supports it, which context and authority govern it, which constraints remain attached, who validated it, who may approve it and how the resulting decision can be challenged.\n\nThe model changes the unit of AI assurance from the model output to the governed decision path. A machine output is not accepted because it is fluent, relevant or confident. It enters as a candidate object and becomes eligible for institutional use only when the identity, evidence, context, authority, constraint and validation predicates required by the applicable policy are satisfied. Eligibility does not itself create a decision. An authorized decision owner remains responsible for approval, rejection, qualification, escalation or refusal.\n\nEvidence-Governed Intelligence consists of eight operational layers: ingestion; entity identity; claim separation; evidence and provenance; relationships and context; constraints and authority; validation and human decision; and audit, correction and contestability.\n\nThe model separates entities from their representations and governs material claims independently of document containers. It preserves source-to-decision provenance, prevents silent context transfer, carries material constraints through transformation, contains conflicting claims until authorized adjudication and records correction as a new governed state rather than erasing prior history.\n\nThe institutional assurance architecture distinguishes the control plane from the processing plane. The control plane maintains identity, policy, authority, access, versions and audit state. The processing plane performs permitted retrieval, extraction, transformation, generation and validation. The processing plane cannot silently rewrite control-plane authority, policy or historical state.\n\nThe paper defines an AI-output verification sequence covering entity identity, evidence coverage, provenance continuity, context compatibility, authority validation, constraint handling, contradiction review and accountable human approval where required. Output states include candidate, qualified candidate, validated under scope, rejected, indeterminate and superseded.\n\nApplication patterns are provided for legal and investigative work, government, family offices, critical infrastructure and formal mathematics. A synthetic family-office investment case demonstrates how identity confusion, jurisdictional mismatch, authority conflict and lost constraints can convert a persuasive AI recommendation into a rejected candidate and a governed conditional decision.\n\nEvaluation metrics include verification time, evidence coverage, provenance completeness, entity-resolution error, contradiction discovery, material-change latency, constraint retention, decision traceability, contestability resolution, unauthorized transfer and the quality of human intervention. Every metric must be defined against a declared baseline and bounded decision domain.\n\nThree implementation pathways provide a controlled adoption progression. A Geometry Intelligence Audit identifies structural gaps in evidence, authority and decision continuity. An Evidence Architecture Pilot applies the model to one bounded decision domain and compares it with the baseline process. A Sovereign Geometry Intelligence Deployment moves validated capability into controlled production under named governance, security, evidence-custody, monitoring, rollback and retirement arrangements.\n\nEvidence-Governed Intelligence complements established AI risk-management, management-system and regulatory frameworks by providing a structural assurance layer between model capability and institutional decision. It does not replace legal, engineering, investment, security or policy expertise. Instead, it makes the evidence, relationships, context, constraints and authority surrounding institutional judgment inspectable, correctable and governable.\n\nThe framework is implementation-neutral. Evaluation can proceed through declared state, evidence records, controlled cases, observable outcomes and retained decision records without mandatory disclosure of proprietary coordinates, source code, model parameters, internal reasoning traces or security-sensitive implementation details.\n\nThis is Paper 5 of the Geometry Intelligence Foundational Series. Paper 1 defines Geometry Intelligence. Paper 2 formalizes governed relational transformation. Paper 3 defines black-box Machine Entity Comprehension conformance. Paper 4 defines optional sovereign federation deployment. Paper 5 applies those foundations to institutional AI assurance and adoption.\n\nThe operating progression is:\n\nGeometry Intelligence Audit → Evidence Architecture Pilot → Sovereign Geometry Intelligence Deployment\n\nAI capability becomes institutional trust when evidence, context, authority and decision accountability remain connected.\n\nFinal Manuscript 1.0. Published by KTS Global on 16 August 2026. Licensed under Creative Commons Attribution 4.0 International.",
      "descriptionType": "Abstract"
    }
  ],
  "relatedIdentifiers": [
    {
      "relationType": "IsSupplementTo",
      "relatedIdentifier": "10.5281/ZENODO.21907367",
      "resourceTypeGeneral": "Text",
      "relatedIdentifierType": "DOI"
    },
    {
      "relationType": "References",
      "relatedIdentifier": "10.5281/ZENODO.21907450",
      "resourceTypeGeneral": "Text",
      "relatedIdentifierType": "DOI"
    },
    {
      "relationType": "References",
      "relatedIdentifier": "10.5281/ZENODO.21907520",
      "resourceTypeGeneral": "Text",
      "relatedIdentifierType": "DOI"
    },
    {
      "relationType": "References",
      "relatedIdentifier": "10.5281/ZENODO.21907588",
      "resourceTypeGeneral": "Text",
      "relatedIdentifierType": "DOI"
    },
    {
      "relationType": "IsVersionOf",
      "relatedIdentifier": "10.5281/zenodo.21962424",
      "relatedIdentifierType": "DOI"
    }
  ],
  "schemaVersion": "http://datacite.org/schema/kernel-4",
  "providerId": "cern",
  "clientId": "cern.zenodo",
  "agency": "datacite",
  "state": "findable"
}