genai-security-crosswalk 2.0.0

package/data/entries/DSGAI07.json

@@ -0,0 +1,680 @@
+{
+  "id": "DSGAI07",
+  "name": "Data Governance and Lifecycle",
+  "source_list": "DSGAI-2026",
+  "version": "2026-Q1",
+  "severity": "High",
+  "aivss_score": null,
+  "audience": [
+    "red-teamer",
+    "security-engineer",
+    "ciso",
+    "compliance",
+    "ml-engineer",
+    "ot-engineer",
+    "auditor",
+    "developer",
+    "data-engineer"
+  ],
+  "mappings": [
+    {
+      "framework": "MITRE ATLAS",
+      "control_id": "AML.T0035",
+      "control_name": "Exfiltrate via ML Inference API",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Ungoverned derived assets (embeddings, caches) accessible through inference API without classification controls"
+    },
+    {
+      "framework": "MITRE ATLAS",
+      "control_id": "AML.T0024.000",
+      "control_name": "Membership Inference",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Ungoverned training data status — adversary determines what sensitive data is in scope without classification barriers"
+    },
+    {
+      "framework": "MITRE ATLAS",
+      "control_id": "AML.T0057",
+      "control_name": "Data from Information Repositories",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Ungoverned embedding stores and agent memory databases accessed without access controls"
+    },
+    {
+      "framework": "NIST AI RMF 1.0",
+      "control_id": "GV-1.6",
+      "control_name": "Policies for data privacy",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Data governance policy extended to all GenAI-derived assets — embeddings, caches, memory, telemetry"
+    },
+    {
+      "framework": "NIST AI RMF 1.0",
+      "control_id": "MP-1.6",
+      "control_name": "Context establishment",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "GenAI data assets inventoried and mapped — source through derivation chain documented"
+    },
+    {
+      "framework": "NIST AI RMF 1.0",
+      "control_id": "MS-3.3",
+      "control_name": "Data quality",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Data quality and governance controls measured across full GenAI data lifecycle"
+    },
+    {
+      "framework": "NIST AI RMF 1.0",
+      "control_id": "MG-3.2",
+      "control_name": "Residual risk",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Residual governance risk from ungoverned derived assets documented and treated"
+    },
+    {
+      "framework": "EU AI Act",
+      "control_id": "High-risk AI training data subject to governance — relevant, representative, privacy-preserving",
+      "control_name": "Art. 10 — Data and data governance",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Data governance policy covering the full AI data lifecycle is an Art. 10 compliance requirement"
+    },
+    {
+      "framework": "EU AI Act",
+      "control_id": "Documented quality management system including data handling procedures",
+      "control_name": "Art. 17 — Quality management",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Data lifecycle procedures — classification, retention, deletion — documented in quality management"
+    },
+    {
+      "framework": "EU AI Act",
+      "control_id": "GPAI providers maintain technical documentation including training data governance",
+      "control_name": "Art. 53(1)(a) — GPAI documentation",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Full data governance documentation for GPAI training data is a binding Art. 53 obligation"
+    },
+    {
+      "framework": "ISO/IEC 27001:2022",
+      "control_id": "A.5.9",
+      "control_name": "Inventory of assets",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "All GenAI data assets inventoried — training data, embeddings, caches, agent memory, logs"
+    },
+    {
+      "framework": "ISO/IEC 27001:2022",
+      "control_id": "A.5.12",
+      "control_name": "Classification of information",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Classification extended to GenAI-derived assets — embeddings inherit source classification"
+    },
+    {
+      "framework": "ISO/IEC 27001:2022",
+      "control_id": "A.5.13",
+      "control_name": "Labelling of information",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Classification labels propagate through the full GenAI data lifecycle"
+    },
+    {
+      "framework": "ISO/IEC 27001:2022",
+      "control_id": "A.8.10",
+      "control_name": "Information deletion",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Deletion and erasure obligations enforced across all derived assets — embeddings, caches, backups"
+    },
+    {
+      "framework": "ISO/IEC 42001:2023",
+      "control_id": "Data — acquisition",
+      "control_name": "A.7.2",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Foundational"
+    },
+    {
+      "framework": "ISO/IEC 42001:2023",
+      "control_id": "Data — preparation",
+      "control_name": "A.7.3",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Hardening"
+    },
+    {
+      "framework": "ISO/IEC 42001:2023",
+      "control_id": "Policies",
+      "control_name": "A.2.2",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Foundational"
+    },
+    {
+      "framework": "ISO/IEC 42001:2023",
+      "control_id": "Planning — risk",
+      "control_name": "Cl.6.1",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Hardening"
+    },
+    {
+      "framework": "CIS Controls v8.1",
+      "control_id": "CIS 3",
+      "control_name": "3.2 — Establish data inventory",
+      "tier": "Foundational",
+      "scope": "Both"
+    },
+    {
+      "framework": "CIS Controls v8.1",
+      "control_id": "CIS 3",
+      "control_name": "3.3 — Configure data access control lists",
+      "tier": "Foundational",
+      "scope": "Both"
+    },
+    {
+      "framework": "CIS Controls v8.1",
+      "control_id": "CIS 3",
+      "control_name": "3.11 — Encrypt sensitive data at rest",
+      "tier": "Foundational",
+      "scope": "Both"
+    },
+    {
+      "framework": "CIS Controls v8.1",
+      "control_id": "CIS 8",
+      "control_name": "8.3 — Ensure adequate audit log storage",
+      "tier": "Foundational",
+      "scope": "Both"
+    },
+    {
+      "framework": "OWASP ASVS 4.0.3",
+      "control_id": "V8 Data Protection",
+      "control_name": "V8.1.4 — Sensitive data minimisation",
+      "tier": "Foundational",
+      "scope": "Both"
+    },
+    {
+      "framework": "OWASP ASVS 4.0.3",
+      "control_id": "V8 Data Protection",
+      "control_name": "V8.3.4 — Sensitive data identified and classified",
+      "tier": "Foundational",
+      "scope": "Both"
+    },
+    {
+      "framework": "OWASP ASVS 4.0.3",
+      "control_id": "V4 Access Control",
+      "control_name": "V4.1.5 — Attribute-based access control",
+      "tier": "Foundational",
+      "scope": "Both"
+    },
+    {
+      "framework": "ISA/IEC 62443",
+      "control_id": "SR 4.1",
+      "control_name": "Data confidentiality",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "OT-derived GenAI assets (embeddings, caches) classified and protected — not treated as general IT data"
+    },
+    {
+      "framework": "ISA/IEC 62443",
+      "control_id": "SR 1.2",
+      "control_name": "Human user authentication",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Access controls on all GenAI-derived OT data assets — same authentication requirements as source data"
+    },
+    {
+      "framework": "ISA/IEC 62443",
+      "control_id": "Supplier security requirements",
+      "control_name": "62443-2-4",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "GenAI vendors handling OT-derived assets subject to 62443-2-4 programme"
+    },
+    {
+      "framework": "NIST SP 800-82 Rev 3",
+      "control_id": "ICS vulnerabilities",
+      "control_name": "§5.3",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Excessive OT data access is a documented vulnerability"
+    },
+    {
+      "framework": "NIST SP 800-82 Rev 3",
+      "control_id": "Risk assessment",
+      "control_name": "§6.2",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Data access scope in OT risk assessment"
+    },
+    {
+      "framework": "NIST SP 800-82 Rev 3",
+      "control_id": "Secure architecture",
+      "control_name": "§7.1",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Least privilege enforced at OT data boundary"
+    },
+    {
+      "framework": "NIST CSF 2.0",
+      "control_id": "GV.OC-01",
+      "control_name": "Organisational Context",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Data governance policy extended to all GenAI-derived assets — embeddings, caches, agent memory"
+    },
+    {
+      "framework": "NIST CSF 2.0",
+      "control_id": "ID.AM-08",
+      "control_name": "Asset Management",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "GenAI data assets inventoried — training datasets, embeddings, RAG stores, agent memory, telemetry logs"
+    },
+    {
+      "framework": "NIST CSF 2.0",
+      "control_id": "PR.DS-01",
+      "control_name": "Data Security",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "All GenAI data assets protected per classification — derived assets inherit source classification"
+    },
+    {
+      "framework": "NIST CSF 2.0",
+      "control_id": "DE.CM-09",
+      "control_name": "Continuous Monitoring",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Monitoring for ungoverned data assets — new GenAI deployments creating untracked data flows detected"
+    },
+    {
+      "framework": "SOC 2",
+      "control_id": "Policy identifies all confidential information in GenAI scope — including derived assets (embeddings, caches, traces)",
+      "control_name": "C1.1 — Confidentiality policy",
+      "tier": "Foundational",
+      "scope": "Both"
+    },
+    {
+      "framework": "SOC 2",
+      "control_id": "Personal information lifecycle management covers GenAI-derived assets — embeddings inherit source data obligations",
+      "control_name": "P4.1 — Privacy information lifecycle",
+      "tier": "Foundational",
+      "scope": "Both"
+    },
+    {
+      "framework": "SOC 2",
+      "control_id": "Data lifecycle procedures documented — classification propagation, retention schedules, deletion procedures",
+      "control_name": "CC5.2 — Control activities",
+      "tier": "Foundational",
+      "scope": "Both"
+    },
+    {
+      "framework": "SOC 2",
+      "control_id": "Ungoverned derived asset risks identified in assessment — embeddings, caches, agent memory outside governance scope",
+      "control_name": "CC3.2 — Risk assessment",
+      "tier": "Foundational",
+      "scope": "Both"
+    },
+    {
+      "framework": "PCI DSS v4.0",
+      "control_id": "Req 3.1.1",
+      "control_name": "Account data inventory",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "All CHD locations documented — derived GenAI assets (embeddings, caches) included in data inventory"
+    },
+    {
+      "framework": "PCI DSS v4.0",
+      "control_id": "Req 3.2.1",
+      "control_name": "Data flow documentation",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Data flow diagrams include all GenAI pipeline paths — source through embedding, retrieval, generation, logging"
+    },
+    {
+      "framework": "PCI DSS v4.0",
+      "control_id": "Req 9.4.6",
+      "control_name": "Media disposal",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Derived GenAI assets containing CHD disposed of securely — deletion verified per Req 9.4 requirements"
+    },
+    {
+      "framework": "PCI DSS v4.0",
+      "control_id": "Req 12.3.2",
+      "control_name": "Targeted risk analysis",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Targeted risk analysis for ungoverned GenAI-derived assets — scope, protection, lifecycle documented"
+    },
+    {
+      "framework": "ENISA Multilayer Framework",
+      "control_id": "L2",
+      "control_name": "Governance and Risk (GOV)",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Comprehensive data governance policy for all GenAI assets — classification, ownership, retention, deletion, and unlearning capability documented"
+    },
+    {
+      "framework": "ENISA Multilayer Framework",
+      "control_id": "L2",
+      "control_name": "Data and Model Security (DMS)",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Data lifecycle controls for training corpora, embeddings, and RAG stores — retention periods enforced, deletion verified"
+    },
+    {
+      "framework": "ENISA Multilayer Framework",
+      "control_id": "MON",
+      "control_name": "Monitoring and Detection",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Data asset inventory monitored for stale or unclassified assets — alerts on policy violations"
+    },
+    {
+      "framework": "ENISA Multilayer Framework",
+      "control_id": "L1",
+      "control_name": "General ICT — Data Protection",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Classified data protected per classification level — access control enforced at storage layer"
+    },
+    {
+      "framework": "OWASP SAMM v2.0",
+      "control_id": "D-SA",
+      "control_name": "Design / Security Architecture",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Define minimum data scope for each system component"
+    },
+    {
+      "framework": "OWASP SAMM v2.0",
+      "control_id": "G-SM",
+      "control_name": "Governance / Strategy & Metrics",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Formalise data access review as part of security programme"
+    },
+    {
+      "framework": "OWASP SAMM v2.0",
+      "control_id": "V-AA",
+      "control_name": "Verification / Architecture Assessment",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Periodic review of declared vs actual data access per component"
+    },
+    {
+      "framework": "OWASP SAMM v2.0",
+      "control_id": "O-OM",
+      "control_name": "Operations / Operational Management",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Alert when component requests data outside declared scope"
+    },
+    {
+      "framework": "OWASP SAMM v2.0",
+      "control_id": "G-PC",
+      "control_name": "Governance / Policy & Compliance",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Documented policy requiring data access justification"
+    },
+    {
+      "framework": "CWE/CVE",
+      "control_id": "CWE-359",
+      "control_name": "CWE-359",
+      "tier": "Foundational",
+      "scope": "Both",
+      "url": "https://cwe.mitre.org/data/definitions/359.html"
+    },
+    {
+      "framework": "CWE/CVE",
+      "control_id": "CWE-213",
+      "control_name": "CWE-213",
+      "tier": "Foundational",
+      "scope": "Both",
+      "url": "https://cwe.mitre.org/data/definitions/213.html"
+    },
+    {
+      "framework": "MAESTRO",
+      "control_id": "L2",
+      "control_name": "Data Operations",
+      "tier": "Foundational",
+      "scope": "Both"
+    },
+    {
+      "framework": "MAESTRO",
+      "control_id": "L6",
+      "control_name": "Security & Compliance",
+      "tier": "Foundational",
+      "scope": "Both"
+    },
+    {
+      "framework": "MAESTRO",
+      "control_id": "L5",
+      "control_name": "Evaluation & Observability",
+      "tier": "Foundational",
+      "scope": "Both"
+    },
+    {
+      "framework": "AIUC-1",
+      "control_id": "A",
+      "control_name": "Data & Privacy domain",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Foundational"
+    },
+    {
+      "framework": "AIUC-1",
+      "control_id": "B006",
+      "control_name": "Prevent unauthorized AI actions",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Foundational"
+    },
+    {
+      "framework": "AIUC-1",
+      "control_id": "B007",
+      "control_name": "Third-party permission controls",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Hardening"
+    },
+    {
+      "framework": "AIUC-1",
+      "control_id": "E",
+      "control_name": "Audit trails and logging",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Foundational"
+    },
+    {
+      "framework": "OWASP NHI Top 10",
+      "control_id": "GenAI system service account has access to more data than declared function requires",
+      "control_name": "NHI-5 Over-Privileged NHI",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Audit and reduce credential scope per system component"
+    },
+    {
+      "framework": "OWASP NHI Top 10",
+      "control_id": "Long-lived credentials maintain excessive access indefinitely",
+      "control_name": "NHI-7 Long-Lived Credentials",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Rotate or replace with short-lived tokens"
+    },
+    {
+      "framework": "OWASP NHI Top 10",
+      "control_id": "Same data access credential used across multiple GenAI system functions",
+      "control_name": "NHI-9 NHI Reuse",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Separate credentials per function"
+    },
+    {
+      "framework": "NIST SP 800-218A",
+      "control_id": "PW.1.1-PS",
+      "control_name": "Define security requirements — data aggregation controls",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Define security requirements governing how datasets may be combined for AI training and inference; require aggregation impact assessments before merging datasets"
+    },
+    {
+      "framework": "NIST SP 800-218A",
+      "control_id": "PW.2.1-PS",
+      "control_name": "Design software — aggregation-aware data architecture",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Design AI data pipelines with aggregation controls that prevent combination of datasets whose joint sensitivity exceeds authorised classification levels"
+    },
+    {
+      "framework": "FedRAMP",
+      "control_id": "AC-6",
+      "control_name": "Least Privilege — data aggregation restrictions",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Restrict data aggregation scope to minimum necessary; enforce controls preventing combination of datasets that create higher-sensitivity aggregates"
+    },
+    {
+      "framework": "FedRAMP",
+      "control_id": "SC-7",
+      "control_name": "Boundary Protection — data flow control",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Enforce boundary protection on data flows between systems; prevent uncontrolled data aggregation across security boundaries"
+    },
+    {
+      "framework": "FedRAMP",
+      "control_id": "PM-9",
+      "control_name": "Risk Management Strategy — aggregation risk",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Include data aggregation risk in AI risk management strategy; assess combined sensitivity of aggregated datasets"
+    },
+    {
+      "framework": "DORA",
+      "control_id": "Art. 5–7",
+      "control_name": "ICT Risk Management — aggregation risk governance",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Include data aggregation risk in ICT risk management; assess combined sensitivity of aggregated financial datasets used in AI systems"
+    },
+    {
+      "framework": "DORA",
+      "control_id": "Art. 9",
+      "control_name": "Protection and Prevention — aggregation controls",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Implement controls preventing uncontrolled data aggregation; enforce restrictions on combining datasets that create higher-sensitivity aggregates"
+    },
+    {
+      "framework": "DORA",
+      "control_id": "Art. 10",
+      "control_name": "Detection — aggregation anomaly detection",
+      "tier": "Foundational",
+      "scope": "Both",
+      "notes": "Monitor for excessive data aggregation patterns; alert on dataset combinations exceeding defined sensitivity thresholds"
+    }
+  ],
+  "tools": [
+    {
+      "name": "Apache Atlas",
+      "type": "open-source",
+      "url": "https://atlas.apache.org"
+    },
+    {
+      "name": "OpenMetadata",
+      "type": "open-source",
+      "url": "https://open-metadata.org"
+    },
+    {
+      "name": "Collibra",
+      "type": "commercial",
+      "url": "https://www.collibra.com"
+    },
+    {
+      "name": "Amundsen",
+      "type": "open-source",
+      "url": "https://github.com/amundsen-io/amundsen"
+    },
+    {
+      "name": "ARX Data Anonymization",
+      "type": "open-source",
+      "url": "https://arx.deidentifier.org"
+    },
+    {
+      "name": "OpenDP",
+      "type": "open-source",
+      "url": "https://opendp.org"
+    },
+    {
+      "name": "Google Differential Privacy",
+      "type": "open-source",
+      "url": "https://github.com/google/differential-privacy"
+    },
+    {
+      "name": "Presidio",
+      "type": "open-source",
+      "url": "https://github.com/microsoft/presidio"
+    },
+    {
+      "name": "Microsoft Presidio",
+      "type": "open-source",
+      "url": "https://github.com/microsoft/presidio"
+    },
+    {
+      "name": "Open Policy Agent",
+      "type": "open-source",
+      "url": "https://www.openpolicyagent.org"
+    }
+  ],
+  "incidents": [
+    {
+      "name": "Samsung employees leak source code and meeting notes via ChatGPT",
+      "url": "https://github.com/emmanuelgjr/GenAI-Security-Crosswalk/blob/main/data/incidents.json",
+      "year": 2023,
+      "incident_id": "INC-001"
+    },
+    {
+      "name": "GitHub Copilot reproduces verbatim licensed code and embedded secrets",
+      "url": "https://github.com/emmanuelgjr/GenAI-Security-Crosswalk/blob/main/data/incidents.json",
+      "year": 2023,
+      "incident_id": "INC-008"
+    },
+    {
+      "name": "Uber ML platform data lineage audit — fragmented provenance across 30+ feature stores",
+      "url": "https://github.com/emmanuelgjr/GenAI-Security-Crosswalk/blob/main/data/incidents.json",
+      "year": 2024,
+      "incident_id": "INC-042"
+    }
+  ],
+  "crossrefs": {
+    "dsgai_2026": [
+      "DSGAI08",
+      "DSGAI01"
+    ],
+    "llm_top10": [
+      "LLM02",
+      "LLM06",
+      "LLM03",
+      "LLM08"
+    ],
+    "agentic_top10": [
+      "ASI03",
+      "ASI02",
+      "ASI06"
+    ]
+  },
+  "changelog": [
+    {
+      "date": "2026-03-27",
+      "version": "1.0.0",
+      "change": "Initial entry — generated from GenAI Security Crosswalk v1.5.1 mapping files",
+      "author": "emmanuelgjr"
+    }
+  ]
+}