# Sample Chunks Index

Quick reference guide to the 20 chunks included in this sample.

## Governance Pack (4 chunks)

| Chunk ID | Title | Source | Type |
|----------|-------|--------|------|
| eu_ai_act_article_006 | Classification rules for high-risk AI systems | EU AI Act Regulation (EU) 2024/1689 | Article |
| nist_rmf_1_1 | NIST AI Risk Management Framework Foundation | NIST AI RMF | Framework intro |
| nist_600_1_govern_1_1 | Governance functions for AI systems | NIST AI 600-1 | Standard section |
| oecd_ai_dd_step_1_1 | Initial AI due diligence process | OECD AI Due Diligence | Process step |

## Security Pack (4 chunks)

| Chunk ID | Title | Source | Type |
|----------|-------|--------|------|
| owasp_llm01_definition | Prompt Injection vulnerability definition | OWASP LLM Top 10 v2.0 | Vulnerability |
| ir8596_govern | Secure AI development governance | NIST IR 8596 | Guidance |
| atlas_aml_m0000 | Reconnaissance baseline tactics | MITRE ATL&CK for ML | Tactic taxonomy |
| cisa_best_practices_001 | AI data security best practices | CISA Secure AI Guidance | Best practice |

## Privacy Pack (4 chunks)

| Chunk ID | Title | Source | Type |
|----------|-------|--------|------|
| EU-AIA-PRIV-ART10-001 | Privacy requirements for AI systems | EU AI Act Article 10 | Regulation |
| CA-SB53-DEF-001 | California AI definitions and scope | California SB-53 | State law |
| EDPB-JO-2026-BG-001 | Algorithmic discrimination and bias | EDPB-EDPS Joint Opinion | Regulatory opinion |
| CISA-SBOM-2025-001 | Software bill of materials for AI transparency | CISA SBOM 2025 | Supply chain standard |

## Financial Services Pack (4 chunks)

| Chunk ID | Title | Source | Type |
|----------|-------|--------|------|
| SR-11-7-001 | Model Risk Management for banks | Federal Reserve SR 11-7 | Supervisory guidance |
| OCC-MRM-001 | Banking oversight of AI model risk | OCC Model Risk Management | Banking guidance |
| SEC-EXAM-2026-001 | Investment adviser AI examination procedures | SEC Exam Procedures 2026 | Regulatory exam guide |
| TREAS-AI-FS-001 | AI governance for financial services | Treasury AI in Financial Services | Agency guidance |

## Healthcare Pack (4 chunks)

| Chunk ID | Title | Source | Type |
|----------|-------|--------|------|
| FDA-SAMD-001 | AI/ML in Software as Medical Device strategy | FDA SaMD AI/ML Action Plan | Strategic action plan |
| FDA-MLT-001 | Machine learning transparency and accountability | FDA ML Transparency Guidance | Regulatory guidance |
| HHS-AI-STR-001 | U.S. healthcare AI strategy and governance | HHS AI Strategy Framework | Strategic framework |
| EMA-AI-RP-001 | European regulatory perspective on healthcare AI | EMA AI Reflection Paper | Reflection paper |

## How to Use This Index

1. **Find chunks by domain**: Scroll to your area of interest (governance, security, privacy, finserv, healthcare).
2. **Understand chunk purpose**: The "Type" column shows what kind of regulatory content this represents.
3. **Locate files**: Each chunk_id maps to two files in sample_chunks/{domain}/:
   - {chunk_id}.md (plain text content)
   - {chunk_id}_metadata.json (structured metadata)
4. **Follow cross-references**: Open any _metadata.json file and check the `cross_references` field to find related chunks in other domains.

## Key Facts

- **Total chunks**: 20
- **Total files**: 40 (20 .md + 20 .json)
- **Domains covered**: 5 (governance, security, privacy, finserv, healthcare)
- **Chunks per domain**: 4
- **Source documents**: 13 authoritative regulatory sources
- **Jurisdictions**: EU, US Federal, California, international (OECD, MITRE, OWASP)

## Metadata Highlights

Each chunk's _metadata.json includes:

- **cross_references**: Links to 3-5 related chunks in other domains
- **enforcement_dates**: Timeline of regulatory requirements (where applicable)
- **ai_actors**: Roles affected by this regulation (provider, deployer, regulator, auditor, etc.)
- **compliance_obligation**: Boolean flag for binding vs. guidance
- **keywords**: Controlled vocabulary for semantic search

Example query using metadata:
```
Find all chunks with compliance_obligation=true AND risk_tier=high_risk AND jurisdiction=European Union
Result: EU AI Act Article 6 and related high-risk system requirements
```

## Next Steps

1. Read the main README.md for schema documentation and licensing.
2. Examine a few chunk files to understand content depth and metadata richness.
3. Load _metadata.json files into your vector database to test performance.
4. Use cross-references to build multi-domain compliance queries.
5. When ready, purchase the full pack(s) for your regulatory area.

---

**Document version**: 1.0.0
**Last updated**: March 29, 2026
**Sample set**: Curated for quality, diversity, and cross-regulatory relevance