Principles
Aligning your AI Council's principles to international standards and frameworks.
Why Principles Matter
Principles provide the normative foundation for council decisions. When a novel case arises that no policy anticipated, the council falls back on principles. They also communicate the organization's commitments externally.
Recommended Principle Sources
The strongest principles frameworks to draw from are:
| Framework | Key Emphases |
|---|---|
| OECD AI Principles (updated May 2024) | Human rights, transparency, robustness, security, safety, accountability |
| UNESCO Recommendation on AI Ethics | Human rights, dignity, fairness, transparency, human oversight |
| Council of Europe Framework Convention | Human rights, democracy, rule of law |
| NIST AI RMF | Govern, Map, Measure, Manage (lifecycle governance) |
| ISO/IEC 42001 | Management system approach, continuous improvement |
Model Principles Set
We recommend adopting 6–8 principles. More than that becomes difficult to remember and apply. Here is a model set:
1. Human Oversight
AI systems support human decision-making. Humans remain accountable for outcomes, and meaningful human oversight is maintained proportionate to risk.
2. Fairness and Non-Discrimination
AI systems are designed, tested, and monitored to avoid unfair bias and discrimination against individuals or groups.
3. Transparency and Explainability
People affected by AI systems can understand that AI is being used, how it influences decisions, and how to seek recourse.
4. Safety and Security
AI systems are robust, secure, and resilient. Security is addressed throughout the lifecycle, including adversarial threats specific to AI.
5. Privacy and Data Governance
AI systems respect privacy rights and are built on data that is lawfully collected, appropriately governed, and fit for purpose.
6. Accountability
Roles and responsibilities for AI governance are clearly defined. Decisions are documented. The organization can demonstrate compliance.
7. Societal and Environmental Wellbeing
The broader impacts of AI systems on society and the environment are considered and, where possible, positive outcomes are pursued.
Using Principles in Practice
Principles are not wall art. They should be:
- Referenced in the charter as the council's normative foundation
- Embedded in intake forms: require teams to self-assess against each principle
- Used in review discussions: structure council deliberation around principles
- Cited in decision records: explain which principles informed each decision
- Published externally: communicate commitments to customers, regulators, and the public