Threat+Constellation
Live— UTC

Execution · NO — NOT EXECUTED

Threat Observatory / Experimental Research Module

When Authority Becomes the Vulnerability

Institutional Dependencies, Decision Latency, and the Limits of Automated Cyber Defense

AI-assisted vulnerability discovery and exploitation may operate faster than institutional decision-making.

The critical governance problem is not merely how quickly vulnerabilities are discovered or patched.

It is whether organizations can authorize, verify, execute, and reverse defensive actions safely within the available response window.

Institutional vulnerability may also be shaped by authority structure and dependency — an exploratory question, not an established empirical conclusion.

Research Status

Version
0.7
Status
Experimental
Evidence Mode
Source-backed / Synthetic / Unknown
Operational Authorization
NOT PROVIDED
Human Review
REQUIRED
Execution
NONE IN v0.7 — NOT EXECUTED

Exploratory question: Is institutional vulnerability determined only by technical exposure, or also by the structure of authority and dependency? This is not an established empirical conclusion.

Research principles

  • Discovery is not authority.
  • Prediction is not evidence.
  • Recommendation is not execution.
  • An exploit attempt is not proof of compromise.
  • Approval is not sufficient without scope, verification, and operational constraints.
  • Public knowledge is not organizational awareness.
  • Delay is not proof of negligence.
  • Knowing that a risk exists is not the same as having the authority, resources, or institutional capacity to resolve it.
  • Dependency is not proof of failure.
  • Technical capability is not formal authority.
  • Absence of evidence is not evidence of inaction.
  • Observed response time is not necessarily total organizational response time.
  • Disclosure is not organizational awareness.
  • Patch availability is not patch deployment.
  • Operational recovery is not necessarily verified recovery.
  • Source credibility does not eliminate the need for claim-level verification.
  • Chronology is not causality.
  • Transmission is not receipt.
  • Receipt is not necessarily awareness.
  • Action is not proof of authorization.
  • Restoration is not verified recovery.
  • Missing evidence is not institutional failure.
  • Traceability is not execution authority.
  • Traceability is not accountability.
  • Knowledge is not authority.
  • Authority is not operational capability.
  • An available patch is not necessarily an available institutional action.
  • Later knowledge must not be projected backward into earlier decisions.
  • Absence of evidence is not evidence of institutional inaction.
  • Responsibility cannot be inferred from chronology alone.
  • Research reviewability is not a finding of legal liability.
  • Possible is not available.
  • Available is not feasible.
  • Feasible is not preferable.
  • Preferable in hindsight is not proven preferable at the time.
  • An unobserved outcome cannot be directly verified.
  • Counterfactual analysis is not a negligence finding.
  • Technical feasibility is not execution authority.
  • Agreement is not truth.
  • Consensus is not independent verification.
  • Unknown is a valid research outcome.
  • Disagreement is evidence of interpretive variation, not necessarily reviewer failure.
  • Methodological reliability is not execution authority.

These statements are conceptual research principles, not empirical facts and not proven conclusions.

02 · Discovery Clock

Machine-speed events

Intervals use only timestamps with established provenance. Missing values remain NOT ESTABLISHED. A date is not a time of day. Negative intervals are valid and labelled INVERTED rather than silently corrected. Date-bounded intervals are not confirmed measurements.

  • Discovery to vendor notification

    discoveryAt → vendorNotificationAt · INCOMPLETE

    NOT ESTABLISHED

  • Discovery to public disclosure

    discoveryAt → publicDisclosureAt · INCOMPLETE

    NOT ESTABLISHED

  • Discovery to patch availability

    discoveryAt → patchAvailableAt · INCOMPLETE

    NOT ESTABLISHED

  • First observed exploitation relative to disclosure

    firstObservedExploitationAt → publicDisclosureAt · mixed precision

    ≈4d 0h (date precision · time of day NOT ESTABLISHED)

  • Disclosure to patch availability

    publicDisclosureAt → patchAvailableAt · mixed precision

    ≈0 (date precision · time of day NOT ESTABLISHED)

AWS reports an exploitation attempt on 17 October 2026, before the 21 October public window. That is an attempt against a honeypot, not confirmed production compromise, and it is not treated as proof that attackers used AI.

03 · Institutional Response Clock

Institutional response events

Public record

  • Public disclosure to organizational detection

    publicDisclosureAt → organizationalDetectionAt · INCOMPLETE

    NOT ESTABLISHED

  • Detection to authorization

    organizationalDetectionAt → decisionAuthorizedAt · INCOMPLETE

    NOT ESTABLISHED

  • Authorization to mitigation start

    decisionAuthorizedAt → mitigationStartedAt · INCOMPLETE

    NOT ESTABLISHED

  • Mitigation start to completion

    mitigationStartedAt → mitigationCompletedAt · INCOMPLETE

    NOT ESTABLISHED

  • Mitigation completion to verified restoration

    mitigationCompletedAt → serviceRestoredAt · INCOMPLETE

    NOT ESTABLISHED

Total response · NOT ESTABLISHED

SYNTHETIC — NOT REAL-WORLD OBSERVATION

  • Public disclosure to organizational detection

    publicDisclosureAt → organizationalDetectionAt · SYNTHETIC

    ≈1d 19h (SYNTHETIC · not an observed metric)

  • Detection to authorization

    organizationalDetectionAt → decisionAuthorizedAt · SYNTHETIC

    ≈1d 7h (SYNTHETIC · not an observed metric)

  • Authorization to mitigation start

    decisionAuthorizedAt → mitigationStartedAt · SYNTHETIC

    ≈2h 30m (SYNTHETIC · not an observed metric)

  • Mitigation start to completion

    mitigationStartedAt → mitigationCompletedAt · SYNTHETIC

    ≈7h 30m (SYNTHETIC · not an observed metric)

  • Mitigation completion to verified restoration

    mitigationCompletedAt → serviceRestoredAt · SYNTHETIC

    ≈4h 0m (SYNTHETIC · not an observed metric)

Decision decision-latency remains labelled inference. It does not become an observed fact.

04 · Authority Surface

Least privilege for defensive actions

A high-confidence recommendation is not permission. This table is an assessment of requested versus granted authority for the public PHP record.

ActionGrantedEnvironmentApprovalReversibility
READ_ONLY_SCANgrantedUNSPECIFIEDnot requiredVERIFIED
PATCH_RECOMMENDATIONgrantedUNSPECIFIEDnot requiredVERIFIED
PRODUCTION_DEPLOYMENTnot grantedPRODUCTIONrequiredUNKNOWN
DATA_DESTRUCTIVE_ACTIONnot grantedPRODUCTIONrequiredNOT_REVERSIBLE

05 · Reversibility Assessment

Reversibility is not recoverability

Reversibility status
UNKNOWN
Recovery readiness
UNKNOWN
  • Rollback availability is not VERIFIED.
  • Backup verification is not VERIFIED.
  • Restore testing is not VERIFIED.
  • Approval is not recorded.
  • Audit trail is not available.

06 · Decision Boundary

ALLOW / HOLD / DENY

Evaluation reuses the existing Kosuke Protocol Decision Boundary. v0.2 never executes. Friction and delay do not determine ALLOW, HOLD, or DENY. Recommendation, authorization, and execution status remain separate fields.

Local advisory evaluator · no network · no execution

Proposed defensive action

Boundary recommendation

ALLOW

Recommendation READ_ONLY_SCAN · Authorization ALLOW · Execution NO — NOT_EXECUTED

  • Provenance and evidence are sufficient for authorization.

Reversibility: Rollback status is UNKNOWN. Backup verification is UNKNOWN. Restore testing is UNKNOWN. Backup is not restore.

Advisory only. Record the analysis. Do not execute infrastructure change.

Did any execution occur? NO — NOT EXECUTED

07 · Known Risk Dwell

Confirmed awareness is not public disclosure

Known Risk Dwell, Unresolved Exposure, and Public Disclosure Age remain separate metrics. Public knowledge is not organizational awareness.

PUBLIC_NAMED · VENDOR_KNOWLEDGE · PUBLIC

PHP Project / PHP Foundation

Known Risk Dwell
≈70d–71d (bounded · time of day NOT ESTABLISHED)
Known Risk Dwell: confirmed awareness of a specific risk by a defined institution to a documented disposition. Disposition is not necessarily remediation. Public disclosure is not substituted for awareness.
Disposition
MITIGATION_COMPLETED
Unresolved Exposure
NOT ESTABLISHED
Public Disclosure Age
2d

PUBLIC_ANONYMIZED · UNKNOWN · PUBLIC

Unnamed operator (no organization-specific evidence)

Known Risk Dwell
NOT ESTABLISHED
Known Risk Dwell: confirmed awareness of a specific risk by a defined institution to a documented disposition. Disposition is not necessarily remediation. Public disclosure is not substituted for awareness.
Disposition
NOT ESTABLISHED
Unresolved Exposure
NOT ESTABLISHED
Public Disclosure Age
2d

SYNTHETIC · ORGANIZATION_CONFIRMED_KNOWLEDGE · SYNTHETIC

SYNTHETIC operator — research fixture (not a real organization)

Known Risk Dwell
≈2d–3d (bounded · time of day NOT ESTABLISHED) · SYNTHETIC · not an observed metric
Known Risk Dwell: confirmed awareness of a specific risk by a defined institution to a documented disposition. Disposition is not necessarily remediation. Public disclosure is not substituted for awareness.
Disposition
MITIGATION_COMPLETED
Unresolved Exposure
NOT ESTABLISHED
Public Disclosure Age
2d · SYNTHETIC · not an observed metric

08 · Institutional Decision Friction

Why a known risk can remain unresolved

Documented reasons, source-stated reasons, research hypotheses, and unknown causes remain separate. Elapsed time is not a cause. See methodology.

  • Risk awareness to acknowledgement

    RISK_SIGNAL_RECEIVED → RISK_ACKNOWLEDGED · INCOMPLETE

    NOT ESTABLISHED

  • Acknowledgement to assessment

    RISK_ACKNOWLEDGED → IMPACT_ASSESSMENT_STARTED · INCOMPLETE

    NOT ESTABLISHED

  • Assessment to decision request

    IMPACT_ASSESSMENT_COMPLETED → DECISION_REQUESTED · INCOMPLETE

    NOT ESTABLISHED

  • Decision request to authorization

    DECISION_REQUESTED → DECISION_AUTHORIZED · INCOMPLETE

    NOT ESTABLISHED

  • Authorization to mitigation start

    DECISION_AUTHORIZED → MITIGATION_STARTED · INCOMPLETE

    NOT ESTABLISHED

  • Mitigation start to completion

    MITIGATION_STARTED → MITIGATION_COMPLETED · date precision

    ≈71d 0h (date precision · time of day NOT ESTABLISHED)

Source-stated / documented

  • AUTHORITY_AMBIGUITY · SOURCE_STATED

    PHP Foundation states that a consensus-based security-team process stalled and restarted work on a fix over months. This is a SOURCE_STATED account of the PHP Project, not a finding about unrelated organizations, and not a negligence verdict.

  • RESOURCE_CONSTRAINT · SOURCE_STATED

    PHP Foundation attributes delay in part to lack of resources and few engineers with deep engine knowledge. SOURCE_STATED. Elapsed time alone is not used as proof of this factor.

  • COMPATIBILITY_RISK · SOURCE_STATED

    PHP Foundation states competing fix approaches risked compatibility breaks, new denial-of-service, or third-party extension changes.

  • COORDINATED_DISCLOSURE_DEPENDENCY · SOURCE_STATED

    Release timing was moved after a leak so distributors could package a fix. Chainguard is stated to have published patched images before the embargo lift. SOURCE_STATED.

Research hypotheses

None recorded.

Unknown causes

  • UNKNOWN · UNKNOWN_CAUSE

    Causes of any operator-side delay at unnamed organizations are UNKNOWN. Public knowledge of the CVE is not evidence that a given website owner knew, deferred, or was negligent.

Governance questions · research fields, not permissions

  • Who owns the risk?

    PHP security team / PHP Foundation for upstream php-src. Operator ownership for deployments is NOT ESTABLISHED. · STATED

  • Who may authorize mitigation?

    NOT ESTABLISHED · VERIFY

  • Who validates the evidence?

    NOT ESTABLISHED · VERIFY

  • Who approves production changes?

    NOT ESTABLISHED · VERIFY

  • Who determines acceptable downtime?

    NOT ESTABLISHED · VERIFY

  • Who verifies recovery?

    NOT ESTABLISHED · VERIFY

  • Who may reopen a risk decision?

    NOT ESTABLISHED · VERIFY

Documented reasons, source-stated reasons, research hypotheses, and unknown causes remain separate. Elapsed time is not a cause. See methodology.

SYNTHETIC — NOT REAL-WORLD OBSERVATION

  • Risk awareness to acknowledgement

    RISK_SIGNAL_RECEIVED → RISK_ACKNOWLEDGED · SYNTHETIC · INCOMPLETE

    NOT ESTABLISHED

  • Acknowledgement to assessment

    RISK_ACKNOWLEDGED → IMPACT_ASSESSMENT_STARTED · SYNTHETIC · INCOMPLETE

    NOT ESTABLISHED

  • Assessment to decision request

    IMPACT_ASSESSMENT_COMPLETED → DECISION_REQUESTED · SYNTHETIC · INCOMPLETE

    NOT ESTABLISHED

  • Decision request to authorization

    DECISION_REQUESTED → DECISION_AUTHORIZED · SYNTHETIC

    ≈20h 0m (SYNTHETIC · not an observed metric)

  • Authorization to mitigation start

    DECISION_AUTHORIZED → MITIGATION_STARTED · SYNTHETIC · INCOMPLETE

    NOT ESTABLISHED

  • Mitigation start to completion

    MITIGATION_STARTED → MITIGATION_COMPLETED · SYNTHETIC · INCOMPLETE

    NOT ESTABLISHED

Source-stated / documented

None recorded.

Research hypotheses

  • AUTHORITY_AMBIGUITY · RESEARCH_HYPOTHESIS · hypothesis

    SYNTHETIC hypothesis: change-board membership was unclear. Not observed. Not negligence.

  • TESTING_REQUIREMENT · RESEARCH_HYPOTHESIS · hypothesis

    SYNTHETIC hypothesis: staging restore tests were required before production. Not observed.

  • SERVICE_CONTINUITY_CONCERN · RESEARCH_HYPOTHESIS · hypothesis

    SYNTHETIC hypothesis: internet-facing checkout could not accept an untested restart window.

Unknown causes

  • UNKNOWN · UNKNOWN_CAUSE

    SYNTHETIC: additional causes remain UNKNOWN.

Governance questions · research fields, not permissions

  • Who owns the risk?

    NOT ESTABLISHED · VERIFY

  • Who may authorize mitigation?

    NOT ESTABLISHED · VERIFY

  • Who validates the evidence?

    NOT ESTABLISHED · VERIFY

  • Who approves production changes?

    NOT ESTABLISHED · VERIFY

  • Who determines acceptable downtime?

    NOT ESTABLISHED · VERIFY

  • Who verifies recovery?

    NOT ESTABLISHED · VERIFY

  • Who may reopen a risk decision?

    NOT ESTABLISHED · VERIFY

09 · Defender's Window

Warning time is signed, bounded, and incomplete

Attempted exploitation, confirmed exploitation, and confirmed compromise remain distinct. A honeypot attempt is not evidence that a named operator was exposed.

Public Defender's Window
−≈4d (date precision · time of day NOT ESTABLISHED) (warning after observed attempt)
Warning: public disclosure 2026-10-21T14:00:00.000Z (datetime). Threat: AWS MadPot attempt 2026-10-17 (date only). Attack evidence ATTEMPTED_EXPLOITATION.
Organizational Defender's Window
NOT ESTABLISHED
No organization-specific warning or environment-scoped threat event is on the public record.

10 · Synthetic Scenario Explorer

Local research fixture

Inputs stay in the browser. They are SYNTHETIC. They cannot authorize production change. Execution remains NOT EXECUTED.

SYNTHETIC · local only · not persisted · NOT EXECUTED

Documented friction factors (synthetic)

Dwell ≈2d–3d (bounded · time of day NOT ESTABLISHED) · SYNTHETIC · not an observed metric

HOLD

Recommendation PRODUCTION_DEPLOYMENT · Execution NO — NOT_EXECUTED

Evidence completeness PARTIAL. Friction does not authorize action. Synthetic data cannot issue production permissions.

Reversibility: Rollback status is UNKNOWN. Backup verification is UNKNOWN. Restore testing is UNKNOWN. Backup is not restore.

11 · Case Studies

PHP 2026 — Vulnerability Discovery and Institutional Response

Public record of CVE-2026-109618 and CVE-2026-109619 (PHP2Shell). Facts below are limited to independently retrieved primary and secondary sources. Organizational response times are not part of this record. Honeypot attempts are not production compromise.

Open the evidence register →

12 · Methodology

Missing timestamps stay unknown. Inferences stay inferences. Synthetic clocks stay labelled synthetic.

Read methodology →

17 · Empirical Evidence

When Institutional Awareness Becomes Institutional Action

An Evidence-Based Study of Vulnerability Disclosure, Organizational Response, and Decision Latency

When does documented institutional awareness become documented institutional action? Candidate historical records remain labeled. Absence of public evidence is not proof of inaction.

Open the empirical evidence layer →

18 · Decision Trace

The Missing Decision Links

Which connections between institutional awareness, authorization, and action can be established from publicly available evidence? Traceability is not accountability.

Open the decision trace →

19 · Institutional Accountability

Before Responsibility Can Be Assigned

Institutional Accountability, Available Authority, and the Evidence of Choice

What must be established before an institution can reasonably be held accountable for a decision, delay, omission, or security response? Research reviewability is not a finding of legal liability.

Open the accountability boundary →

20 · Institutional Counterfactuals

The Choices That Were Never Taken

Institutional Counterfactuals, Historical Constraints, and the Limits of Retrospective Judgment

At a particular historical decision point, what alternative actions were actually available to an institution, and what can legitimately be inferred about their possible consequences? A counterfactual is not an observed historical event. Feasibility is not execution authority.

Open the counterfactual boundary →

21 · Independent Evidence Validation

Can Evidence Boundaries Be Reproduced?

Inter-Rater Reliability, Epistemic Disagreement, and the Reproducibility of Institutional Accountability Assessment

Can independent researchers applying the same evidence definitions reach reproducible classifications of institutional knowledge, authority, feasibility, and accountability reviewability? Agreement is not truth. Methodological reliability is not execution authority.

Open independent validation →

Connected observatories