AICardSummary — Model Card Generator (NIST AI RMF + EU AI Act Annex IV)
Interactive model-card generator. Inputs: model name + intended use + out-of-scope use + training data + eval results + bias analysis + mitigations + governance. Live preview + HTML/Markdown export. Conforms to NIST AI RMF + EU AI Act Annex IV (high-risk AI system documentation, mandatory from 2026).
What it is
The companion to PromptGovernance + ModelEvalPanel. Where PromptGovernance tracks operational prompts and ModelEvalPanel scores accuracy, AICardSummary generates the public-facing artifact every EU AI Act high-risk system needs by 2026.
What’s in it
- Live-updating model-card generator with 9 input fields
- Pre-filled with realistic fraud-scoring example (XGBoost + 142 features)
- Outputs include: intended use, out-of-scope use (the often-missed honesty signal), training-data composition, eval results (precision/recall/F1/calibration/subgroup performance), bias analysis, mitigations + safeguards
- HTML + Markdown export
- EU AI Act risk classification (high-risk Annex III / limited-risk transparency / minimal-risk)
Why this shape
NIST AI RMF 1.0 (Jan 2023) made documentation a foundational practice. EU AI Act Annex IV (effective for high-risk systems from Aug 2026) makes a specific shape of model card mandatory. ISO/IEC 42001 §8.4 codifies it. Mitchell et al. “Model Cards for Model Reporting” (2019) is the academic origin.
How it ships
Single HTML file, ~10KB. Zero dependencies + zero persistence (session-only). 9 inputs × live preview + HTML/MD export in 100 lines of vanilla JavaScript.