AI Goverance
As artificial intelligence (AI) technology becomes more widely applied across business operations, Nan Pao upholds the principles of safety, transparency, trustworthiness, and human oversight in advancing Responsible AI governance, ensuring that the adoption and application of AI technology complies with information security, regulatory compliance, ethical principles, and sustainability goals. The Company has established an AI Governance Policy applicable to the Company and all Group subsidiaries in Taiwan and abroad, serving as the basis for AI adoption, use, management, and oversight, with the governance mechanism reviewed on an ongoing basis to reduce potential risk and improve the benefits of AI application.
To implement AI governance, the Company has established the following AI management mechanisms to ensure AI applications are properly monitored and managed:
| Area | Description |
|---|---|
| 1. AI Risk Assessment and Adoption Management | AI systems, models, or tools must undergo a risk assessment prior to adoption, comprehensively considering the purpose of the application, data type, information security, regulatory compliance, and potential impact on stakeholders. Applications involving personal data, confidential information, or high-risk matters must complete an assessment of necessity, reasonableness, and compliance prior to adoption. Access to sensitive or high-risk AI capabilities, including facial recognition and surveillance tools, is restricted to authorized personnel and subject to prior approval. |
| 2. AI Governance and Human Oversight | AI serves only as a decision-support tool; material or high-risk matters must still undergo human review by accountable personnel. AI-generated content and AI-supported decision outputs are clearly labeled or logged as appropriate to the use case, and model accuracy, stability, and bias risk are reviewed regularly to continuously improve model governance quality. |
| 3. System Controls and Continuous Improvement | Appropriate system controls and manual verification mechanisms are established for AI-automated processes, ensuring AI output is checked as necessary before being imported into core systems, reducing operational risk. AI application effectiveness is continuously tracked as a basis for optimizing management measures and improving governance maturity. The Company also monitors deployed AI models for drift, performance degradation, and other deviations over time and takes corrective action when issues are identified. |
| 4. Sustainable Application, Third-Party Management, and Capability Building | The Company encourages applying AI to process optimization, quality management, energy management, supply chain risk management, and occupational safety and health, to improve operating efficiency and sustainability performance. When adopting third-party AI services, the Company assesses their information security, data protection, model governance, and service stability, and continues to provide training on AI ethics, information security, personal data protection, and generative AI use to strengthen employees' capability to use AI responsibly. |
To strengthen employees' capability to use AI responsibly and effectively, the Company organized the following AI-related training courses:
| Course | Cumulative Training Hours (hrs) | Participants |
|---|---|---|
| Everyday Work Tasks, Taught Once with AI | 6 | 1 |
| [Remote Digital Software Skills Workshop] From Documents to Video — Complete Guide to AI Workplace Applications | 18 | 1 |
| [Online] ChatGPT / AI Tools for Efficient Office Administration and Documentation | 3 | 1 |
| AI Course Series: NotebookLM + n8n | 84 | 84 |
| AI Course Series: The Art of Effective Prompting | 60 | 60 |
| AI-Powered Work Skills for the Trading Industry | 36 | 3 |
| In-Person Workshop: Essential AI Knowledge for Non-IT HR Professionals + Organization-Centered Workforce Inventory — AI-Era Workforce Inventory Series 1 | 6 | 1 |
| Easy Guide to Canva AI Presentation Design | 6 | 1 |
The Company completed the AI Governance Policy in 2026, establishing a governance framework for AI adoption, use, and management, and completed the related management procedures and division of responsibilities, forming the foundation for advancing AI governance and risk management going forward. Management indicators and an implementation-tracking mechanism will be progressively established to continuously review AI application effectiveness and governance maturity.