# XVIII. Security

### Part 18 — Handling, Safety, Security, and Data Protection 

*(Binding objective: prevent information hazard, rights harm, targeting cues, market harm, and integrity failure while preserving contestability and correctionability. Safety constraints override convenience, growth, or publicity.)*

#### 18.1 Purpose, Scope, and Binding Safety Posture

18.1.1 **Safety as first-order objective.** The Platform is an evidence commons; measurement and publication can cause harm; safety controls are non-optional for any opted-in action that creates or distributes artifacts.\
18.1.2 **Non-executing perimeter alignment.** Handling/safety controls must reinforce the non-executing boundary (no command, no dispatch, no enforcement, no procurement steering, no operational intelligence tasking).\
18.1.3 **System linkages.** This Part governs and constrains Quality System (Part 17), due process/appeals (Part 22), and Registry posture (Part 24), including stop-the-line holds and corrections discipline.\
18.1.4 **Voluntary baseline with opt-in obligations.** No one is compelled to participate; obligations attach only when a participant elects handling, enters controlled lanes, accepts a role, or receives controlled materials.\
18.1.5 **No silent relaxation.** Safety/handling protections are non-regressive: rule weakening requires record-valid change control, explicit rationale, contestation window, and effective date; “quiet downgrades” are prohibited.

***

#### 18.2 Handling Class Framework

18.2.1 **Canonical handling classes.** Public-Safe / Controlled / Restricted / Special (event-sensitive or legally constrained).\
18.2.2 **Mandatory handling election.** Every artifact, workspace, convening, CCell/CERT, and release must have a recorded handling class election (defaulting rules apply only until election is made).\
18.2.3 **Default handling rules.** Public-Safe is default; elevation is mandatory when content includes sensitive datasets, critical infrastructure dependencies, security vulnerabilities, market-sensitive information, identity risk, or rights-impacting material.\
18.2.4 **Distinct concepts rule.** Handling class (distribution risk) is distinct from confidentiality (legal duty), IP (ownership/licensing), and privacy (personal data); overlap may occur but labels are not interchangeable.\
18.2.5 **Handling survival obligations.** Post-exit obligations attach to Controlled/Restricted materials: no redistribution, deletion/return attestations where applicable, and continued correction duties for authored outputs.

***

#### 18.3 Distribution Controls and “Need-to-Know” Discipline

18.3.1 **Distribution logs.** Mandatory for Controlled/Restricted and for any Public-Safe summary derived from controlled evidence when recipients must be tracked; minimum fields include: artifact ID/version, recipient role marker, access grant basis, expiry, revocation pointer, and delivery channel class.\
18.3.2 **Copy-control defaults.** Controlled lanes default to no-forward / no-copy / no-print / no-public-paste unless explicitly elected by record with rationale.\
18.3.3 **Watermarking defaults.** Per-recipient watermarking is default for Controlled/Restricted; watermark drills and leak-response hooks must exist for each lane.\
18.3.4 **Time-boxed access and revocation-by-record.** Controlled access is time-boxed; expiration is default; revocation is record-valid with reason codes; prior recipients must be reconcilable for correction distribution.\
18.3.5 **Secure channels and access minimization.** Controlled distribution requires secured channels, short-lived links, least privilege, and access logging; attachments are minimized or replaced with controlled pointers where feasible.\
18.3.6 **Metadata minimization.** Calendar-safe titles, attendee list minimization, neutral meeting descriptions, and obfuscated location/time windows where required to reduce targeting and deanonymization risk.

***

#### 18.4 Information Hazard Doctrine

18.4.1 **Dual-use screening baseline.** All outputs and convenings have dual-use screening; escalation triggers include actionable threat enablement, CI targeting cues, exploit detail, or destabilizing market/public narratives.\
18.4.2 **Critical infrastructure targeting-cue prohibition.** Publication must be abstraction-first; operationally actionable facility/location/system diagrams, dependency maps, or precise vulnerabilities are prohibited unless tightly restricted and lawful.\
18.4.3 **Market sensitivity controls.** Content that could move credit/insurance markets, reveal MNPI, expose systemic dependencies, or cause destabilization must use delayed release, controlled annexing, and strict reliance bounds.\
18.4.4 **Panic amplification controls.** Narrative safety is mandatory: avoid speculation, avoid rumor propagation, avoid sensational framing; safe summaries prioritize uncertainty disclosure and correction paths.\
18.4.5 **Exploit disclosure boundaries.** No “how-to attack” content; vulnerability reporting routes to responsible disclosure lanes and/or controlled safety gate decisions.\
18.4.6 **Redaction/partition/staged release.** Default pattern: public-safe summary first, controlled annex second; partition sensitive details; delay where harm risk is time-dependent.

***

#### 18.5 Safety Gates for Publication and Convenings

18.5.1 **Stewardship safety gate.** Safety gate roles decide outcomes: approve / hold / restrict / redact / delay / partition / require cross-council review; decisions are record-valid and reason-coded.\
18.5.2 **Stop-the-line holds.** Stop-the-line is permitted only for safety/integrity; holds require mandatory expiry, reopen criteria, and recorded basis; indefinite silent holds are prohibited.\
18.5.3 **Handling-aware review lanes.** Review routing depends on handling class; cross-council triggers apply for CI/cyber/AI coupling, rights impacts, and market sensitivity.\
18.5.4 **Session readiness gates.** Clinics/labs/briefings require pre-session readiness: docket, limitations, handling election, distribution plan, COI log, recording election, and correction clock.\
18.5.5 **Post-release monitoring.** Monitor for misuse indicators, harm signals, and misrepresentation; trigger corrections, takedowns, badge revocations, or handling escalations by record.

***

#### 18.6 Security Baseline for Members, Workspaces, and Artifacts

18.6.1 **Account security by tier.** MFA required for elevated access; device posture guidance for Controlled/Restricted lanes; secure recovery and emergency lock policies.\
18.6.2 **Least privilege and scoping.** Role-marker privileges are time-boxed and revocable; privilege grants are recorded with scope and expiry.\
18.6.3 **Secure workspace rules.** Controlled rooms require access logging, restricted export controls, controlled pointers, and (where needed) secure sandboxes/TEEs for compute-to-data.\
18.6.4 **Supply-chain hygiene.** Shared tooling must follow signed releases where applicable, SBOM posture, and change-control discipline; dependency risks are tracked in release notes.\
18.6.5 **Threat modeling baseline.** High-risk domains (CI/cyber/AI/bio/market-sensitive finance interfaces) require threat modeling and red-team format checks before dissemination.\
18.6.6 **Insider risk and privilege creep detection.** Periodic access reviews, anomaly detection, rotation, and audit sampling for privilege abuse and “quiet expansions.”

***

#### 18.7 Incident Handling

18.7.1 **Incident categories.** Security incidents (breach/leak/compromise), safety incidents (harmful dissemination/dual-use failure), conduct incidents (harassment/doxxing), integrity incidents (misrepresentation/fraud/collusion).\
18.7.2 **Reporting channels and protected participation.** Confidential/anonymous reporting options as lawful; anti-retaliation rules; protected challenger pathways.\
18.7.3 **Triage clocks and escalation thresholds.** Time-boxed triage and escalation; coordination limits reaffirmed (no operational command); reason-coded decisions.\
18.7.4 **Evidence preservation.** Registry-based evidence preservation (logs, artifact hashes, distribution logs, decision records) as chain-of-custody equivalent for governance purposes.\
18.7.5 **Communications integrity.** No speculation; safe summaries; uncertainty disclosure; embargo/delay controls where needed; misquote response hooks.\
18.7.6 **After-action records.** Mandatory after-action record for material incidents: what changed, what held, what corrected, what withheld, and which controls were updated.

***

#### 18.8 CERT Handling Rules

18.8.1 **Inheritance rule.** CERT inherits the highest relevant handling class of its inputs unless explicitly elevated; downgrades are prohibited absent formal redaction/partition and safety gate approval.\
18.8.2 **Formation handling fields.** CERT formation record must include: handling class, scope, lawful basis notes, liaison posture, distribution plan, expiry, and after-action obligations.\
18.8.3 **Distribution logging + expiry.** CERT outputs in Controlled/Restricted require distribution logs and default expiry; revocation-by-record must be possible.\
18.8.4 **Artifact segregation.** Separate public-safe summary from controlled annex; never embed sensitive details in public-safe outputs.\
18.8.5 **Safe liaison posture.** CERT may route disclosures and evidence; it cannot dispatch, command, or enforce; no implied authority language is permitted.\
18.8.6 **Dissolution and archive controls.** CERT dissolves by default; archival must preserve corrections linkage and handling survival obligations.

***

#### 18.9 Data Protection Baseline

18.9.1 **Minimization and purpose limitation.** Collect/store only what is necessary for evidence work and governance integrity; prohibit “nice-to-have” identity/data collection.\
18.9.2 **PII/PHI default prohibition.** PII/PHI is prohibited by default; where unavoidable, require lawful basis documentation, minimization, elevated handling, access logs, and explicit retention limits.\
18.9.3 **Sensitive categories.** Youth/vulnerable persons, protected participation, and high-risk groups require additional safeguarding, reduced attribution, and restrictive handling.\
18.9.4 **Sovereign compatibility.** Support compute-to-data and sovereign data zone posture; avoid bulk export; allow in-jurisdiction processing patterns.\
18.9.5 **Retention and deletion.** Retention schedules by handling class; deletion/de-identification requests honored within record-integrity constraints (e.g., restrict access, de-identify, preserve record pointers).\
18.9.6 **Cross-border constraints.** Export controls, sanctions, and local law constraints govern access, sharing, and tooling; abstention and restriction are default when uncertain.

***

#### 18.10 Identity Minimization Cell

18.10.1 **Role markers ≠ identity.** Default participation via role markers; pseudonym support is permitted; identity is not required unless lawful/handling conditions mandate it.\
18.10.2 **Limited identity cell.** A minimal identity cell (records/register + COI/ethics + security/handling) maintains mapping under strict access logs and two-person rule.\
18.10.3 **Unmasking criteria.** Unmasking only for lawful necessity, safety necessity, or due process; never for curiosity, prestige, or influence disputes.\
18.10.4 **Two-person rule.** Dual authorization required; access is logged, time-boxed, and reason-coded; disclosures are minimized.\
18.10.5 **Identity breach response.** Rapid containment, notification posture as required by law/handling, and remediation record; attribution defaults to off during and after breaches.

***

#### 18.11 Consent, Recording, and Photography Rules

18.11.1 **Room class governs recording.** Recording permissions are room-class dependent and must be elected by record; default is no recording in Controlled/Restricted.\
18.11.2 **Consent and default prohibitions.** Explicit consent required for recording/quotes/photos; default prohibitions apply absent consent; consent is scoped and expiry-bound.\
18.11.3 **Attendee minimization.** Attendee lists are optional; role-marker rosters are preferred; identity lists only where required.\
18.11.4 **Artifact capture rules.** Notes/screenshots inherit handling class; survival obligations attach; prohibited capture in restricted lanes unless explicitly elected.\
18.11.5 **Publication permissions.** Images/quotes require permissioned attribution with expiry; misquote response and correction routing are mandatory.

***

#### 18.12 Third-Party Tools, Integrations, and Data Room Controls

18.12.1 **Approved tooling posture.** Allowlists for Controlled/Restricted lanes; tools must support access logging, least privilege, and export controls.\
18.12.2 **Prohibited tooling.** Tools enabling uncontrolled copy-out, shadow distribution, covert recording, or unlogged access are prohibited for controlled lanes.\
18.12.3 **Controlled data rooms.** Tiered access, watermarking, expiry, audit logs, and compute-to-data controls; bulk extraction is prohibited by default.\
18.12.4 **Vendor neutrality.** Tooling choices and integration recommendations must not steer procurement; no endorsements or shortlists.\
18.12.5 **Exit/portability.** Portability is permitted only without safety regression: equivalent-or-stronger handling; preserve correction links and distribution reconciliation capability.

***

#### 18.13 Training, Competence, and Recertification for Sensitive Lanes

18.13.1 **Handling training by tier.** Handling competence is gated by PoC mapping; elevated lanes require renewed training and attestations.\
18.13.2 **Controlled dissemination competence.** Controlled dissemination privileges require PoC thresholds and demonstrated capability (partition, staged release, leak response).\
18.13.3 **Renewal clocks.** Renewal is time-based and change-triggered (rule changes, incident learnings, tool changes, breach events).\
18.13.4 **Failure modes.** Lapse or breach triggers suspension/downgrade pending retraining; reinstatement requires record-valid requalification.\
18.13.5 **Accessibility accommodations.** Training must provide reasonable accommodations; language/low-bandwidth options are part of safety (to avoid error).

***

#### 18.14 Audit, Testing, and Continuous Improvement

18.14.1 **Distribution log audits.** Sampling audits verify completeness, recipient eligibility, expiry compliance, and correction redistribution reconciliation.\
18.14.2 **Watermark drills.** Periodic watermark/leak drills validate containment, attribution, and takedown mechanics.\
18.14.3 **Access reviews.** Regular privilege reviews detect creep, stale grants, and abnormal patterns; high-impact roles require tighter cadence.\
18.14.4 **Correction-clock performance.** Safety-related correction clocks are measured; repeated misses trigger process changes and potential privilege downgrades.\
18.14.5 **Non-regression reviews.** Any rule change is reviewed for safety regression; regression is prohibited absent emergency patch procedures with expiry and ratification.\
18.14.6 **Rule updates by record.** All updates require impact statement, notice and contestation window, effective dates, and supersession mapping; emergency patches require after-action and ratification.


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.therisk.global/organization/cooperation/nexus-guilds/membership/xviii.-security.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.