Interactive public proof

See what happens before an AI proposal becomes organizational consequence.

Choose a public-safe case. Inspect the authoritative ALLOW, BLOCK, or ESCALATE result, release control, consequence proof, route maturity, controlled-batch posture, and browser-side packet verification.

Explore a governed decisionVerify the public packet

Controlled event-level proof is shown. Production L5 route completeness and non-bypassability require customer-specific integration evidence.

Current public sanitized packetFour controlled synthetic casesData-driven controlled batchBrowser-side SHA-256 and parity checks
1 · Decision Explorer

Choose the movement the AI attempted to cause.

The Explorer displays the packet’s authoritative decision and release fields. Semantic Projection remains explanatory and cannot replace them.

Loading

Loading proof case…

Workflow
—
Attempted movement
—
Potential protected effect
—
Authoritative Decision API result
LOADING
Release authorized
—
Release status
—
Routed to
—
Gate result
—
No-bind / proof status—
Protected effect formed—
Downstream commit—
Authorization token—
Human review required—
Open governance PDFInspect buyer JSONInspect decision receipt
2 · Governed Event Lifecycle

Eight stages connect proposal to consequence and closure.

Every lifecycle value comes from the current public packet or the selected packaged controlled case.

    3 · Controlled Batch Proof

    Four proposed movements. One conservative roll-up.

    The chart, counts, aggregate result, human-attention total, filters, and table are computed from the same controlled-case data.

    Aggregate decision—Any BLOCK controls; otherwise ESCALATE controls; all ALLOW is required for aggregate ALLOW.
    Aggregate release—Authorized only when every case is release-authorized.
    Human attention—Cases requiring review or blocked from release.
    Proof postureControlled batchNot customer production activity or a real-time activity counter.

    Showing all controlled cases.

    CaseAttempted consequenceDecisionReleaseProof
    4 · Browser-Side Verification

    Do not take the packet on faith.

    Fetch the public manifest first, verify listed SHA-256 hashes, and compare canonical decision, release, and no-bind fields. SHA-256 integrity checking is not an external digital signature.

    No files are uploaded. Verification runs in this browser against same-origin public assets.

    Verification resultNot run0 files checked · 0 verified
    • Manifest loadWaiting
    • Listed SHA-256 hashesWaiting
    • Decision parityWaiting
    • Release-status parityWaiting
    • No-bind parityWaiting
    Public capability boundary

    What this Explorer proves—and what it does not claim.

    Shown here

    • Decision API action
    • Release state and route
    • Controlled no-bind evidence
    • Eight-stage lifecycle
    • Controlled batch posture
    • Public packet integrity and parity checks

    Requires customer integration

    • Production endpoint enforcement
    • Production route completeness
    • Fail-closed route tests
    • Replay evidence
    • Reviewer and outcome closure

    Not claimed

    • Universal production L5
    • External digital signing
    • Customer validation
    • Automatic compliance
    • Live model-weight mutation
    Move from public proof to your workflow

    Which AI-proposed movement should OntoGuard govern first?

    Start with 10–25 representative prompts, responses, outputs, or logs. No internal document upload is required for the first assessment.

    Request an AI Output Risk ScanDiscuss a controlled pilotDownload complete public proof kit
    Risk ScanExplore