Hong Kong Ethical AI Principles
Tags: Frameworks
Categorisation
The framework defines 12 Ethical AI Principles, grouped as:
- Performance Principles (foundational):
- Transparency and Explainability – Decision-making processes of AI applications should be explainable to humans clearly and comprehensibly.
- Reliability, Robustness and Security – AI should operate reliably over time using appropriate models and datasets; remain robust (consistent, error-tolerant) and secure against cyber-attacks, compliant with legal/industry requirements.
- Fairness – Treat individuals in similar groups fairly, without favouritism, discrimination, or harm.
- Diversity and Inclusion – Promote inclusion and respect stakeholders’ interests.
- Human Oversight – Calibrate human intervention to the perceived severity of ethical issues.
- Lawfulness and Compliance – Act in accordance with applicable laws, regulations, and regulatory regimes.
- Data Privacy – Uphold rights consistent with the Personal Data (Privacy) Ordinance (PD(P)O) Data Protection Principles (DPP1–DPP5): purpose/manner of collection; accuracy/retention; use of personal data; security; and information to be generally available.
- Safety – Do not compromise physical safety or mental integrity throughout the AI application’s lifetime.
- Accountability – Organisations are responsible for moral implications; clearly identify an accountable party.
- Beneficial AI – Promote the common good.
- Cooperation and Openness – Foster multi-stakeholder, open cooperation within the AI ecosystem.
- Sustainability and Just Transition – Ensure mitigation strategies for societal and environmental impacts.
Role in the Framework
- The Performance Principles underpin effective execution of all other principles (e.g., without reliability/robustness/security, compliance to other principles cannot be verified consistently).
- Principles are operationalised through the AI Practice Guide and validated via the AI Application Impact Assessment (AAIA) at key lifecycle stages.
Derivation
The principles draw on the UN Universal Declaration of Human Rights and Hong Kong ordinances, providing a values-based foundation for ethical AI within local legal contexts.
Sources:- Ethical_AI_Framework_en.pdf
← Back to Knowledge Base