AI Trustworthiness and Ethics Charter

1. Foundations

Trustworthy AI within ARTIFACT is guided by:

• The EU AI Act
• NIST AI Risk Management Framework
• OECD AI Principles

2. Core Principles

1. Transparency — model provenance, documentation, and explainability
2. Accountability — traceable decision pipelines and audit trails
3. Robustness — validated, stress-tested models with uncertainty quantification
4. Fairness — mitigation of dataset or model bias
5. Privacy — data minimisation and secure computation
6. Human oversight — human-in-the-loop evaluation and control

3. Model Documentation

Each model should include:

• Dataset and preprocessing description
• Architecture and training details
• Evaluation metrics (accuracy, ECE, calibration, etc.)
• Limitations and risk assessment

4. Governance

The ARTIFACT Ethics Group periodically reviews compliance and publishes annual trust reports.