Skip to content

Add OWASP AISVS 1.0 mapping across all three source lists (25 → 26 frameworks) #94

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
emmanuelgjr merged 2 commits into from
Sep 14, 2026
Merged

Add OWASP AISVS 1.0 mapping across all three source lists (25 → 26 frameworks) #94

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
1db6fc0
Add OWASP AISVS 1.0 mapping (25 → 26 frameworks)
emmanuelgjr Sep 14, 2026
542fb7e
Map LLM Top 10 2026 to AISVS; merge main
emmanuelgjr Sep 14, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
13 changes: 13 additions & 0 deletions CHANGELOG.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,6 +9,19 @@ Versioning follows [Semantic Versioning](https://semver.org/spec/v2.0.0.html).

## [Unreleased]

### Added

- **OWASP AISVS 1.0** mapped to all three source lists — 25 → 26 frameworks.
`llm-top10/LLM_AISVS.md` (82 rows, LLM01–LLM10), `agentic-top10/Agentic_AISVS.md` (83 rows, ASI01–ASI10) and
`dsgai-2026/DSGAI_AISVS.md` (141 rows, DSGAI01–DSGAI21), plus `data/frameworks/owasp-aisvs.json` holding all 191
AISVS 1.0 requirements (inventory `complete`, verified against upstream `1.0/en`). Authored by Otto Sulin (AISVS
co-lead) in GenAI-Security-Project/GenAI-Data-Security-Initiative#64 and ported here.
- The LLM file was authored against the 2025 list and re-keyed to the LLM Top 10 2026 per `MIGRATION.md`: sections
renumbered and renamed, cross-references renumbered, severities re-baselined, and 8 requirements added for the 2026
scope changes (LLM01 cross-modal 2.2.3/2.2.4, LLM04 artifact provenance 3.1.1/3.1.3, LLM05 fine-tuning subversion
6.1.2/3.5.1, LLM08 hidden context 10.2.4, LLM10 generated code 9.3.7). AISVS has no requirement for scanning
generated code itself; the LLM10 section says so.

Next: npm publish to npmjs.com, custom domain (crosswalk.owasp.org), vendor integration packs, NeMo Guardrails configs.

---
Expand Down
49 changes: 28 additions & 21 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,14 +5,14 @@
<!-- stats:badges -->
[![Version](https://img.shields.io/badge/version-4.0.0-green)](CHANGELOG.md)
[![Source Lists](https://img.shields.io/badge/source%20lists-4-blueviolet)](README.md)
[![Mapping Files](https://img.shields.io/badge/mapping%20files-74-brightgreen)](README.md)
[![Frameworks](https://img.shields.io/badge/frameworks-25-orange)](README.md)
[![Mapping Files](https://img.shields.io/badge/mapping%20files-77-brightgreen)](README.md)
[![Frameworks](https://img.shields.io/badge/frameworks-26-orange)](README.md)
<!-- /stats -->
[![npm](https://img.shields.io/npm/v/genai-security-crosswalk?color=red&label=npm)](https://www.npmjs.com/package/genai-security-crosswalk)

> The most comprehensive publicly available mapping of OWASP GenAI security risks to industry frameworks —
> covering LLM applications, autonomous agentic AI, and GenAI data security across
> **<!-- stats:frameworks-mapped -->25<!-- /stats --> frameworks** and
> **<!-- stats:frameworks-mapped -->26<!-- /stats --> frameworks** and
> **<!-- stats:source-lists -->4<!-- /stats --> OWASP source lists**.

## [Live Web App](https://genai-security-project.github.io/crosswalk/) | [Score Your Coverage](https://genai-security-project.github.io/crosswalk/#/score) | [Explore Entries](https://genai-security-project.github.io/crosswalk/#/explorer) | [View Incidents](https://genai-security-project.github.io/crosswalk/#/incidents)
Expand All @@ -30,7 +30,7 @@ AI risks to all of them.

**This repo solves that.** Every OWASP GenAI vulnerability
(<!-- stats:entries -->51<!-- /stats --> total) is mapped to specific controls in
<!-- stats:frameworks-mapped -->25<!-- /stats --> industry frameworks.
<!-- stats:frameworks-mapped -->26<!-- /stats --> industry frameworks.
Pick your risk, find your controls.

### 3 ways to use it (pick one)
Expand Down Expand Up @@ -74,14 +74,14 @@ Every file answers one question: **which controls from framework X address vulne
| | |
|---|---|
| **<!-- stats:source-lists -->4<!-- /stats -->** source lists | LLM Top 10 · Agentic Top 10 · DSGAI 2026 · Agentic Skills Top 10 |
| **<!-- stats:frameworks-mapped -->25<!-- /stats -->** frameworks | Compliance · Governance · Threat modeling · Testing · OT/ICS · Identity · Secure SDLC · Financial |
| **<!-- stats:mapping-files -->74<!-- /stats -->** mapping files | Every source list entry × every applicable framework |
| **Mapping review state** | <!-- stats:frameworks-draft -->2 of 25 carry candidate DRAFT rows only — CoSAI · EU AI Act Code of Practice<!-- /stats --> — see [docs/SCHEMA_V2_MIGRATION.md](docs/SCHEMA_V2_MIGRATION.md) |
| **Framework freshness** | <!-- stats:freshness -->2 current · 1 behind upstream · 22 unchecked<!-- /stats --> — see [docs/FRESHNESS_SLA.md](docs/FRESHNESS_SLA.md) |
| **<!-- stats:frameworks-mapped -->26<!-- /stats -->** frameworks | Compliance · Governance · Threat modeling · Testing · OT/ICS · Identity · Secure SDLC · Financial |
| **<!-- stats:mapping-files -->77<!-- /stats -->** mapping files | Every source list entry × every applicable framework |
| **Mapping review state** | <!-- stats:frameworks-draft -->2 of 26 carry candidate DRAFT rows only — CoSAI · EU AI Act Code of Practice<!-- /stats --> — see [docs/SCHEMA_V2_MIGRATION.md](docs/SCHEMA_V2_MIGRATION.md) |
| **Framework freshness** | <!-- stats:freshness -->3 current · 1 behind upstream · 22 unchecked<!-- /stats --> — see [docs/FRESHNESS_SLA.md](docs/FRESHNESS_SLA.md) |
| **21** implementation recipes | Production-ready Python patterns |
| **70+** open-source tools | Catalogued and organised by function |
| **25** eval profiles | Runnable Garak (13) + PyRIT (6) + LAAF (6) tests mapped to OWASP entries |
| **<!-- stats:frameworks-mapped -->25<!-- /stats -->** compliance reports | Per-framework gap assessments auto-generated from data layer (MD, CSV, JSON, OSCAL) |
| **<!-- stats:frameworks-mapped -->26<!-- /stats -->** compliance reports | Per-framework gap assessments auto-generated from data layer (MD, CSV, JSON, OSCAL) |
| **<!-- stats:incidents -->131<!-- /stats -->** documented incidents | Real-world + research incidents with MAESTRO layer attribution (MD, CSV, JSON, STIX 2.1) |
| **LAAF v2.0** | First agentic LPCI red-teaming framework — fully integrated with 6-stage × OWASP crosswalk |

Expand Down Expand Up @@ -114,6 +114,7 @@ All free. All open-source. Built for practitioners.
| [ISO/IEC 42001:2023](https://www.iso.org/standard/81230.html) | ✅ | ✅ | ✅ |
| [CIS Controls v8.1](https://www.cisecurity.org/controls) | ✅ | ✅ | ✅ |
| [OWASP ASVS 4.0.3](https://owasp.org/www-project-application-security-verification-standard/) | ✅ | ✅ | ✅ |
| [OWASP AISVS 1.0](https://github.com/OWASP/AISVS/tree/main/1.0/en) | ✅ | ✅ | ✅ |
| [SOC 2 Trust Services Criteria](https://www.aicpa-cima.com/resources/landing/2017-trust-services-criteria) | ✅ | ✅ | ✅ |
| [PCI DSS v4.0](https://www.pcisecuritystandards.org/document_library/) | ✅ | ✅ | ✅ |
| [ENISA Multilayer Framework](https://www.enisa.europa.eu/publications/multilayer-framework-for-good-cybersecurity-practices-for-ai) | ✅ | ✅ | ✅ |
Expand All @@ -129,7 +130,7 @@ All free. All open-source. Built for practitioners.

## All mapping files

### LLM Top 10 2026 — <!-- stats:frameworks-llm -->25<!-- /stats --> framework mappings
### LLM Top 10 2026 — <!-- stats:frameworks-llm -->26<!-- /stats --> framework mappings

| File | Framework | Standout content |
|---|---|---|
Expand All @@ -156,8 +157,9 @@ All free. All open-source. Built for practitioners.
| [LLM_SP800218A.md](llm-top10/LLM_SP800218A.md) | NIST SP 800-218A | Secure AI SDLC practices — PW/PS/RV practice mapping per LLM entry |
| [LLM_FedRAMP.md](llm-top10/LLM_FedRAMP.md) | FedRAMP | SP 800-53 AI overlay — AC/AU/CA/CM/IA/IR/RA/SA/SC/SI/SR control families |
| [LLM_DORA.md](llm-top10/LLM_DORA.md) | DORA | EU financial sector resilience — Art. 5–45 per LLM entry |
| [LLM_AISVS.md](llm-top10/LLM_AISVS.md) | OWASP AISVS 1.0 | Level 1–3 verification requirements for all 10 entries — injection defenses, output control, agent budgets, cross-modal inputs |

### Agentic Top 10 2026 — <!-- stats:frameworks-agentic -->25<!-- /stats --> framework mappings
### Agentic Top 10 2026 — <!-- stats:frameworks-agentic -->26<!-- /stats --> framework mappings

| File | Framework | Standout content |
|---|---|---|
Expand All @@ -183,11 +185,12 @@ All free. All open-source. Built for practitioners.
| [Agentic_SP800218A.md](agentic-top10/Agentic_SP800218A.md) | NIST SP 800-218A | Secure agentic SDLC — tool access, memory integrity, multi-agent pipeline practices |
| [Agentic_FedRAMP.md](agentic-top10/Agentic_FedRAMP.md) | FedRAMP | Federal agentic AI authorization — agent identity, tool access, cascade controls |
| [Agentic_DORA.md](agentic-top10/Agentic_DORA.md) | DORA | Financial sector agentic resilience — incident reporting, third-party agent risk |
| [Agentic_AISVS.md](agentic-top10/Agentic_AISVS.md) | OWASP AISVS 1.0 | Testable "Verify that" requirements, Levels 1–3, weighted to C9 Orchestration and C10 MCP Security |

> **Also in this folder:** [Agentic_CWE_CVE.md](agentic-top10/Agentic_CWE_CVE.md) — CWE root cause taxonomy,
> confirmed CVEs, full CWE cross-reference index.

### DSGAI 2026 — <!-- stats:frameworks-dsgai -->23<!-- /stats --> framework mappings
### DSGAI 2026 — <!-- stats:frameworks-dsgai -->24<!-- /stats --> framework mappings

| File | Framework | Standout content |
|---|---|---|
Expand All @@ -212,6 +215,7 @@ All free. All open-source. Built for practitioners.
| [DSGAI_SP800218A.md](dsgai-2026/DSGAI_SP800218A.md) | NIST SP 800-218A | Secure GenAI data SDLC — training data protection, data governance, provenance practices |
| [DSGAI_FedRAMP.md](dsgai-2026/DSGAI_FedRAMP.md) | FedRAMP | Federal data security controls — SC-28 data at rest, AU-2 logging, SR supply chain |
| [DSGAI_DORA.md](dsgai-2026/DSGAI_DORA.md) | DORA | Financial data resilience — Art. 8 asset inventory, Art. 12 backup, Art. 28-44 vendor risk |
| [DSGAI_AISVS.md](dsgai-2026/DSGAI_AISVS.md) | OWASP AISVS 1.0 | Level 1–3 verification requirements for all 21 entries — retrieval-time authorization, embedding sanitization, training-data provenance |

### Shared resources

Expand Down Expand Up @@ -244,7 +248,7 @@ crosswalk/
│ ├── index.ts ← Typed API: getEntry, getFramework, searchEntries
│ └── index.test.ts ← 12 smoke tests (Node.js built-in runner)
│
├── llm-top10/ ← LLM01–LLM10 × <!-- stats:frameworks-llm -->25<!-- /stats --> frameworks
├── llm-top10/ ← LLM01–LLM10 × <!-- stats:frameworks-llm -->26<!-- /stats --> frameworks
│ ├── LLM_MITREATLAS.md
│ ├── LLM_NISTAIRMF.md
│ ├── LLM_EUAIAct.md
Expand All @@ -267,9 +271,10 @@ crosswalk/
│ ├── LLM_NHI.md ← Non-Human Identity controls
│ ├── LLM_SP800218A.md ← Secure AI SDLC (SSDF extension)
│ ├── LLM_FedRAMP.md ← US federal cloud AI (SP 800-53 overlay)
│ └── LLM_DORA.md ← EU financial sector resilience
│ ├── LLM_DORA.md ← EU financial sector resilience
│ └── LLM_AISVS.md ← AI Security Verification Standard — testable requirements
│
├── agentic-top10/ ← ASI01–ASI10 × <!-- stats:frameworks-agentic -->25<!-- /stats --> frameworks
├── agentic-top10/ ← ASI01–ASI10 × <!-- stats:frameworks-agentic -->26<!-- /stats --> frameworks
│ ├── Agentic_AIUC1.md
│ ├── Agentic_MITREATLAS.md
│ ├── Agentic_NISTAIRMF.md
Expand All @@ -292,9 +297,10 @@ crosswalk/
│ ├── Agentic_NISTSP80082.md ← OT/ICS — U.S. regulatory alignment
│ ├── Agentic_SP800218A.md ← Secure agentic SDLC
│ ├── Agentic_FedRAMP.md ← Federal agentic AI authorization
│ └── Agentic_DORA.md ← Financial sector agentic resilience
│ ├── Agentic_DORA.md ← Financial sector agentic resilience
│ └── Agentic_AISVS.md ← AI Security Verification Standard — testable requirements
│
├── dsgai-2026/ ← DSGAI01–DSGAI21 × <!-- stats:frameworks-dsgai -->23<!-- /stats --> frameworks
├── dsgai-2026/ ← DSGAI01–DSGAI21 × <!-- stats:frameworks-dsgai -->24<!-- /stats --> frameworks
│ ├── DSGAI_ISO27001.md
│ ├── DSGAI_NISTAIRMF.md
│ ├── DSGAI_EUAIAct.md
Expand All @@ -315,7 +321,8 @@ crosswalk/
│ ├── DSGAI_NHI.md ← Non-Human Identity — data pipeline credentials
│ ├── DSGAI_SP800218A.md ← Secure GenAI data SDLC
│ ├── DSGAI_FedRAMP.md ← Federal data security controls
│ └── DSGAI_DORA.md ← Financial data resilience
│ ├── DSGAI_DORA.md ← Financial data resilience
│ └── DSGAI_AISVS.md ← AI Security Verification Standard — testable requirements
│
├── shared/
│ ├── RECIPES.md ← 21 implementation patterns (Python code)
Expand Down Expand Up @@ -371,7 +378,7 @@ crosswalk/
Generate framework-specific gap assessments from the data layer in seconds:

```bash
node scripts/compliance-report.js # all <!-- stats:frameworks-mapped -->25<!-- /stats --> frameworks → reports/
node scripts/compliance-report.js # all <!-- stats:frameworks-mapped -->26<!-- /stats --> frameworks → reports/
node scripts/compliance-report.js --framework "EU AI Act" # one framework
node scripts/compliance-report.js --format csv # Excel-compatible
node scripts/compliance-report.js --format json # machine-readable
Expand Down Expand Up @@ -432,8 +439,8 @@ No install required. Works on desktop and mobile.
| Page | What it does |
|------|-------------|
| [**Score**](https://genai-security-project.github.io/crosswalk/#/score) | Select your frameworks, see coverage gaps. Upload Garak/PyRIT/LAAF results to validate. Share your score card on LinkedIn. |
| [**Explorer**](https://genai-security-project.github.io/crosswalk/#/explorer) | Search and filter all <!-- stats:entries -->51<!-- /stats --> entries. Click any entry to see controls across all <!-- stats:frameworks-mapped -->25<!-- /stats --> frameworks. |
| [**Frameworks**](https://genai-security-project.github.io/crosswalk/#/frameworks) | Interactive <!-- stats:entries -->51<!-- /stats -->×<!-- stats:frameworks-mapped -->25<!-- /stats --> coverage matrix. Click any cell to see the specific controls mapped. |
| [**Explorer**](https://genai-security-project.github.io/crosswalk/#/explorer) | Search and filter all <!-- stats:entries -->51<!-- /stats --> entries. Click any entry to see controls across all <!-- stats:frameworks-mapped -->26<!-- /stats --> frameworks. |
| [**Frameworks**](https://genai-security-project.github.io/crosswalk/#/frameworks) | Interactive <!-- stats:entries -->51<!-- /stats -->×<!-- stats:frameworks-mapped -->26<!-- /stats --> coverage matrix. Click any cell to see the specific controls mapped. |
| [**Incidents**](https://genai-security-project.github.io/crosswalk/#/incidents) | Browse 50 AI security incidents. Filter by severity, year, MAESTRO layer. Full attribution details. |

**Evidence-based scoring** — three validation tiers:
Expand Down
Loading
Loading