multilevel governance

Originating from political science to describe shared power structures in entities like the European Union, multilevel governance is now applied to complex technology governance, including AI. Its core definition is a dynamic framework where multiple levels of stakeholders—governments (international, national), the private sector (corporations, industry associations), and civil society (NGOs, academia, individuals)—collaborate through continuous interaction and negotiation to manage risks and establish norms. Within a risk management system, it bridges internal policies and external regulations. Unlike top-down regulation or corporate self-regulation, it emphasizes collaborative responsibility. Implementing this framework helps companies meet the principles of the NIST AI Risk Management Framework (AI RMF 1.0), which calls for diverse stakeholder engagement, and aligns with the multi-stakeholder approach of the ISO/IEC 42001 AI management system standard. For enterprises, it means moving beyond mere compliance to proactively shaping a responsible AI ecosystem with regulators and users.

Enterprises can apply multilevel governance to AI risk management through three practical steps: 1. **Stakeholder Mapping and Engagement:** Identify and map all relevant stakeholders, including regulators (e.g., EU authorities, Taiwan's NCC), industry associations, technology suppliers, and end-user groups. Establish formal communication channels, such as regulatory briefings, user feedback surveys, and participation in industry standard-setting workshops. 2. **Integrated Governance Structure Design:** Form an internal, cross-functional AI Ethics and Risk Committee, incorporating legal, technical, product, and compliance teams, as guided by the NIST AI RMF's "Govern" function. This body is responsible for translating external requirements from regulations like the EU AI Act and public expectations into concrete internal policies, development guidelines, and control measures. 3. **Transparency Reporting and Feedback Loops:** Implement a public transparency reporting mechanism to disclose AI system risk assessments, algorithmic decision-making logic, and bias mitigation efforts. For example, a global fintech firm that introduced this framework for its AI lending model saw customer appeals decrease by 20% and achieved a 100% pass rate on regulatory audits within one year.

Taiwan enterprises face three primary challenges when implementing AI multilevel governance: 1. **Fragmented and Evolving Regulations:** Taiwan's core AI legislation is still under development, and existing rules are scattered across multiple ministries (e.g., Ministry of Digital Affairs, Financial Supervisory Commission). This lack of a unified standard creates uncertainty and increases compliance costs. 2. **Resource Constraints for SMEs:** Establishing a dedicated AI governance team and implementing international frameworks like NIST AI RMF or ISO/IEC 42001 requires significant financial investment and specialized talent, posing a major barrier for the small and medium-sized enterprises that dominate Taiwan's economy. 3. **Weak Foundation of Public Trust:** Effective collaboration depends on trust between the government, corporations, and the public. However, widespread concern in Taiwan over personal data privacy and the "black box" nature of algorithms hinders meaningful public engagement and data-sharing initiatives. **Solutions:** Enterprises should adopt a phased approach. Start by forming a virtual, cross-departmental task force. Prioritize adopting a structured framework like ISO/IEC 42001 as an internal starting point. Actively participate in industry associations to collectively provide policy feedback to the government, helping to co-create industry standards and best practices.

Winners Consulting specializes in multilevel governance for Taiwan enterprises, delivering compliant management systems within 90 days. Free consultation: https://winners.com.tw/contact