Hugging Face's logo

Datasets:
Zeezhu
/
grc-security-frameworks

Tasks:
Question Answering
Text Generation
Modalities:
Text
Formats:
parquet
Languages:
English
Size:
1K - 10K
Tags:
cybersecurity
compliance
grc
governance
risk
nist
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
Community
1
grc-security-frameworks / README.md
Zeezhu's picture
Zeezhu
Upload dataset
a33d0f0 verified
preview code
|
raw
history blame contribute delete
7.52 kB
metadata 
license: cc-by-4.0
task_categories:
  - question-answering
  - text-generation
language:
  - en
tags:
  - cybersecurity
  - compliance
  - grc
  - governance
  - risk
  - nist
  - cis-controls
  - cloud-security
size_categories:
  - 1K<n<10K
dataset_info:
  - config_name: alpaca
    features:
      - name: instruction
        dtype: string
      - name: input
        dtype: string
      - name: output
        dtype: string
      - name: source
        dtype: string
      - name: id
        dtype: string
    splits:
      - name: train
        num_bytes: 2801898
        num_examples: 2154
    download_size: 836652
    dataset_size: 2801898
  - config_name: chatml
    features:
      - name: messages
        list:
          - name: role
            dtype: string
          - name: content
            dtype: string
      - name: source
        dtype: string
      - name: id
        dtype: string
    splits:
      - name: train
        num_bytes: 3225039
        num_examples: 2154
    download_size: 851148
    dataset_size: 3225039
configs:
  - config_name: alpaca
    data_files:
      - split: train
        path: alpaca/train-*
  - config_name: chatml
    data_files:
      - split: train
        path: chatml/train-*

GRC Security Frameworks Dataset

A comprehensive dataset for training AI models on Governance, Risk, and Compliance (GRC) frameworks and cybersecurity standards.

Dataset Overview

This dataset contains 3,225 high-quality training examples covering major security and compliance frameworks. It's designed for fine-tuning large language models to become expert GRC assistants.

Covered Frameworks

  • CIS Controls v8.1.2 - 153 safeguards across 18 control families
  • Cloud Controls Matrix (CCM) v4.0.12 - 197 cloud security controls
  • NIST SP 800-53 Rev 5 - 200 security and privacy controls
  • NIST Cybersecurity Framework v2.0 - 22 subcategories across 6 functions
  • NIST AI Risk Management Framework v1.0 - 72 AI governance actions

Dataset Structure

Configurations

The dataset is available in two formats:

1. Alpaca Format (alpaca)

Standard instruction-tuning format with three fields:

{
  "instruction": "What is CIS Control 1.1?",
  "input": "Provide details about this safeguard.",
  "output": "CIS Control 1.1: Establish and Maintain Detailed Enterprise Asset Inventory...",
  "metadata": {
    "source_framework": "CIS Controls v8.1.2",
    "control_id": "1.1",
    "dataset_type": "unified_controls"
  }
}

2. ChatML Format (chatml)

Conversational format with role-based messages:

{
  "messages": [
    {
      "role": "system",
      "content": "You are a GRC compliance expert..."
    },
    {
      "role": "user",
      "content": "What is CIS Control 1.1?"
    },
    {
      "role": "assistant",
      "content": "CIS Control 1.1: Establish and Maintain Detailed Enterprise Asset Inventory..."
    }
  ],
  "metadata": {
    "source_framework": "CIS Controls v8.1.2",
    "control_id": "1.1",
    "dataset_type": "unified_controls"
  }
}

Dataset Splits

  • Alpaca: 3,225 examples

    • Unified Controls: 797 examples
    • Framework Mappings: 2,167 examples
    • Assessment Questions: 261 examples

  • ChatML: 3,072 examples

    • Unified Controls: 644 examples
    • Framework Mappings: 2,167 examples
    • Assessment Questions: 261 examples

Usage

Load with Hugging Face Datasets 

from datasets import load_dataset

# Load Alpaca format
dataset = load_dataset("Zeezhu/grc-security-frameworks", "alpaca")

# Load ChatML format
dataset = load_dataset("Zeezhu/grc-security-frameworks", "chatml")

# Split into train/test
dataset = dataset['train'].train_test_split(test_size=0.1, seed=42)
train_data = dataset['train']
test_data = dataset['test']

Example Training Use Case

This dataset is ideal for:

  • Fine-tuning models for GRC advisory chatbots
  • Training compliance automation systems
  • Building security framework mapping tools
  • Creating assessment question generators
  • Developing control implementation assistants

Dataset Creation

Source Data

Original data extracted from official framework publications:

  • CIS Controls v8.1.2 (JSON)
  • CSA Cloud Controls Matrix v4.0.12 (JSON)
  • NIST SP 800-53 Rev 5 (OSCAL JSON)
  • NIST Cybersecurity Framework v2.0 (JSON)
  • NIST AI RMF Playbook v1.0 (JSON)

Processing Pipeline

  1. Extraction: Parsed official JSON/OSCAL files
  2. Normalization: Unified schema across frameworks
  3. Augmentation: Generated Q&A pairs and mappings
  4. Validation: Quality checks and format verification
  5. Formatting: Converted to Alpaca and ChatML formats

Quality Assurance

  • ✅ 100% format validation
  • ✅ No duplicates across datasets
  • ✅ Consistent metadata tagging
  • ✅ Source attribution for all examples
  • ✅ Manual spot-checks of content accuracy

Content Categories

1. Unified Controls (644-797 examples)

Individual control explanations with:

  • Control ID and title
  • Full description
  • Implementation guidance
  • Asset type relevance
  • Security function mapping

2. Framework Mappings (2,167 examples)

Cross-framework relationships showing:

  • How controls map between frameworks
  • Related controls across standards
  • Equivalent requirements
  • Compliance alignment

3. Assessment Questions (261 examples)

Interview-style questions for:

  • Control implementation verification
  • Compliance gap analysis
  • Audit preparation
  • Risk assessment

Limitations

  • Dataset reflects framework versions as of generation date
  • Does not include proprietary or restricted frameworks
  • Focus on technical controls; limited governance/policy content
  • English language only
  • May not reflect latest framework updates after 2024

Ethical Considerations

  • Intended Use: Educational, compliance automation, GRC advisory systems
  • Misuse Risks: Should not replace professional security audits or legal compliance advice
  • Accuracy: While sourced from official frameworks, always verify critical compliance decisions
  • Bias: Reflects cybersecurity industry standards and may not cover all global regulations

Citation

If you use this dataset, please cite:

@dataset{grc_security_frameworks_2025,
  title={GRC Security Frameworks Dataset},
  author={Zeezhu},
  year={2025},
  publisher={Hugging Face},
  url={https://huggingface.co/datasets/Zeezhu/grc-security-frameworks}
}

License

This dataset is released under CC BY 4.0 (Creative Commons Attribution 4.0 International).

You are free to:

  • Share — copy and redistribute the material
  • Adapt — remix, transform, and build upon the material

Under the following terms:

  • Attribution — You must give appropriate credit

Framework Licenses

Source frameworks retain their original licenses:

  • CIS Controls: CIS Terms of Use
  • NIST publications: Public domain (U.S. Government work)
  • CSA CCM: Creative Commons Attribution 4.0

Contact

For questions, issues, or contributions:

Version History

  • v1.0 (2025): Initial release
    • 3,225 Alpaca examples
    • 3,072 ChatML examples
    • 5 major frameworks covered
    • 644 unified controls
    • 2,167 framework mappings
    • 261 assessment questions

Acknowledgments

Special thanks to:

  • Center for Internet Security (CIS) for CIS Controls
  • Cloud Security Alliance (CSA) for CCM
  • National Institute of Standards and Technology (NIST) for SP 800-53, CSF, and AI RMF
  • The open-source compliance community