Skip to main content
This page combines our Risk Management Policy (methodology) with the Risk Register (the live tracked-risks list). External auditors usually want to see these together. Both are reviewed quarterly and whenever a material change occurs.
Owner: Security Officer · Reviewed by: CTO + CEO at management review · Policy version: 1.0

Purpose

Denialbase’s risk management process:
  1. Identifies risks to the confidentiality, integrity, and availability of information assets.
  2. Assesses likelihood and impact to produce a consistent severity score.
  3. Decides treatment: mitigate, transfer, accept, or avoid.
  4. Tracks treatment actions to completion.
  5. Monitors residual risk and reviews periodically.
This aligns with ISO 27001:2022 clauses 6.1.2 + 6.1.3 (risk assessment + risk treatment) and ISO 27005 (guidance).

Methodology

Scope of assessment

  • All information assets listed in our asset inventory (applications, infrastructure, data, processes).
  • All workforce + vendor relationships.
  • All changes to scope trigger a delta assessment.

Identification

Risks are identified from:
  • Threat intelligence — industry advisories, CVE monitoring, healthcare-specific threat reports.
  • Audit findings — internal and external.
  • Incident reviews — root causes of actual incidents.
  • Vendor changes — new subprocessors, changed terms, vendor breaches.
  • Architecture reviews — new features, new data flows, new integrations.
  • Workforce input — anyone can propose a risk; reviewed at next risk meeting.
  • Annual threat modeling workshop — facilitated STRIDE walkthrough of the full stack.

Assessment — 5×5 matrix

Each identified risk is scored on two dimensions: Likelihood (probability the risk materializes in the next 12 months):
ScoreLabelDefinition
5Very HighAlmost certain — expected within 3 months
4HighLikely — expected within 12 months
3MediumPossible — could happen within 12–24 months
2LowUnlikely — possible but not expected
1Very LowRare — theoretical
Impact (consequence if it does):
ScoreLabelDefinition
5SeverePHI breach affecting >500 individuals; >$1M cost; existential regulatory action
4MajorPHI breach < 500 individuals; material service disruption (>24 hours); significant regulatory inquiry
3ModerateNon-PHI data exposure; service disruption 4–24 hours; minor regulatory notification
2MinorInternal-data exposure; service disruption < 4 hours; localized operational impact
1NegligibleNo customer impact; internal process friction only
Inherent risk score = Likelihood × Impact (range 1–25). Residual risk score = Likelihood × Impact after controls are applied.

Rating bands

ScoreBandResponse
16–25CriticalTreat immediately; no acceptance without CEO + Security Officer sign-off
10–15HighTreat within 90 days; Security Officer sign-off on treatment
5–9MediumTreat within 180 days; documented treatment plan
1–4LowMonitor; treat if cost-effective

Treatment decisions

For each risk above the Low band, choose one:
  • Mitigate — implement controls to reduce likelihood and/or impact.
  • Transfer — contractually shift risk (insurance, BAA indemnification, vendor SLA).
  • Accept — document acceptance with rationale and expiry (maximum 12 months, renewable).
  • Avoid — eliminate the source (e.g. don’t use the vendor, don’t build the feature).
Every risk above Low has a named owner accountable for the treatment action.

Risk acceptance

For risks we choose to accept (i.e. not fully mitigate):
  • Documented in the register with rationale.
  • Expiry — acceptance is time-bound (default 6 months).
  • Compensating controls — any alternative mitigations in place.
  • Approval — Security Officer for Medium/High; CEO + Security Officer for Critical.
  • Review — acceptance revisited at every management review.

Communication

  • Top-5 risks reported to every management review.
  • Critical or High risks with customer-visible implications reported on the Trust Center (this page).
  • Internal risk register (with owner names, ticket links, evidence paths) available to all workforce via Google Drive.

Review cadence

ActivityCadence
Owner updates their entriesMonthly
Full register review with CTO + CEOQuarterly
New-risk intake (any workforce member can propose)Continuous
New-risk triage + classificationWithin 14 days of proposal
External review (part of audit)Annual
Full methodology reviewAnnual

Register (public snapshot — April 2026)

This is a sanitized snapshot. Internal entries include owner names, ticket links, evidence file paths, and private vendor evaluation records. Enterprise customers can request the full internal register under mutual NDA: security@denialbase.com.

Critical (score 16–25)

None currently. All Critical risks from the April 2026 audit review have been moved to active treatment with progress tracked quarterly.

High (score 10–15)

IDRiskCategoryL × ITreatmentTarget closure
R-2026-001Unsigned subprocessor BAAs (GCP, Anthropic, Sentry, SES) create HIPAA compliance gapSupplier / Legal4 × 4 = 16Mitigate — BAA execution in progress; technical PHI-minimization compensating controls in place (encryption, scrubbing, private networking, audit logging)Q3 2026
R-2026-002PHI column-encryption migration incomplete — some PHI columns stored as plaintext at the ORM layer despite CMEK at restConfidentiality3 × 4 = 12Mitigate — migration + backfill in progress (PR #78); compensating controls: CMEK, private VPC, audit loggingQ2 2026
R-2026-003No formal Incident Response Plan with tested runbook — placeholders in current draftOperational3 × 4 = 12Mitigate — IRP formalization + first tabletop exerciseQ2 2026
R-2026-004No DR drill executed — backups in place but restore procedure untested at scaleOperational / Availability3 × 4 = 12Mitigate — schedule first backup-verification + PITR drillQ2 2026
R-2026-005No third-party penetration test — automated SAST + dep scans don’t substitute for adversarial testingSecurity3 × 4 = 12Mitigate — engage CREST-accredited firm; scope external app + infrastructureQ3 2026
R-2026-008No formalized ISP / AUP / policy stack ahead of ISO 27001 auditGovernance3 × 4 = 12Mitigate — this Trust Center publication; internal sign-off by CEOQ2 2026 (in progress on this branch)

Medium (score 5–9)

IDRiskCategoryL × ITreatmentTarget closure
R-2026-006JWT key rotation has dual-secret validation support but no scheduled rotation cadence executed in productionSecurity2 × 3 = 6Mitigate — implementation shipped (PR #76); first rotation drill Q2 2026Q2 2026
R-2026-007GCP organization policies (block SA key creation/upload) gated behind disabled Terraform flagInfrastructure2 × 3 = 6Mitigate — enable in staging first (PR #77); promote to prod after staging confirmsQ2 2026
R-2026-009HR onboarding/offboarding records not consolidated — practices exist but not audit-exportable from one placeHR / Governance3 × 2 = 6Mitigate — consolidate into HR platform with exportable recordsQ3 2026
R-2026-010LLM prompt-injection or data-exfil risk in AI-generated appeals if scrubbing failsAI Safety2 × 3 = 6Mitigate — PHI scrubbing before LLM call; output validation; HITL review for low-confidence; periodic prompt-injection testingOngoing monitoring
R-2026-011Workforce phishing risk — no formal simulation program yetHR / Security2 × 3 = 6Mitigate — adopt phishing simulation platformQ2 2026
R-2026-012Key-person dependency (small team) on CTO and Security OfficerOperational2 × 3 = 6Mitigate — document all CTO-only runbooks; cross-train; external advisor retainer for Security OfficerQ3 2026
R-2026-013No Sigstore/cosign image signing yet — supply chain integrity relies on GCP Artifact Registry access controlSupply chain2 × 3 = 6Mitigate — adopt Sigstore signing + verification at deploy timeQ3 2026
R-2026-014No formal asset inventory — controls assume inventory existsAsset management2 × 2 = 4 → adj. 6 for audit-readinessMitigate — create and maintain asset inventoryQ2 2026
R-2026-015Kaiser-specific integration relies on a single integration point; Kaiser outage impacts a material portion of the bookAvailability2 × 3 = 6Transfer / Mitigate — SLA clauses; fallback for most common flows; diversificationQ4 2026

Low (score 1–4)

Tracked internally; not repeated here. Examples include: ESLint rule coverage gaps, documentation freshness drift, minor UX issues with DR runbook clarity.

Relationship to other documents

Threat modeling ──┐
Audit findings ───┼→ Risk register ──→ Risk treatment plan ──→ Actions (tickets)
Incident reviews ─┤                         │
Change impact ────┘                         ↓
                               Management review → Resource allocation
  • Every audit finding with residual risk becomes a risk register entry.
  • Every risk treatment decision is tracked to closure.
  • Every quarterly review updates the register and carries outputs into the next management review.

Records retained

RecordRetention
Internal risk register (versioned)6 years
Risk acceptance records6 years past expiry
Management review minutes (risks discussed)6 years
External audit reports (risks surfaced)6 years

Changelog

VersionDateMaterial changes
0.12026-04Draft informal register (10 entries)
1.02026-04Formal methodology + 15-entry register with L×I scoring, bands, treatment decisions