Governance

Artificial Intelligence (AI) is reshaping our industry — accelerating innovation, transforming operations, and unlocking levels of precision, efficiency, and sustainability previously beyond reach. As a leading enterprise, UMC has embraced AI as a strategic capability across business and operational functions, from automation and predictive analytics to smart manufacturing.

With this opportunity comes responsibility. AI systems must be designed, deployed, and operated with the same engineering discipline, regulatory rigor, and ethical standards that define our enterprise. Our stakeholders — including customers, employees, suppliers, and regulators — deserve AI that is secure, transparent, fair, reliable, and sustainable.

This Responsible AI Policy codifies our commitments. It is endorsed by the Board of Directors, governed by a dedicated AI Governance function, and aligned with international standards. AI is not a constraint on innovation — it is the foundation of durable, trusted, and sustainable progress. We are committed to that foundation.

Principle 1. Human-Centric

• Human Oversight & Intervention: AI must not be the sole basis for high-stakes decisions; apply tiered oversight by human-in-command / human-on-the-loop / human-in-the-loop.

Principle 2. Privacy by Design

• Data Privacy & Compliance: Enforce lawful, necessary and proportionate processing, data minimization, purpose limitation, role-based access, and end-to-end data provenance across the full AI system lifecycle.

Principle 3. Security by Design

• Cybersecurity & Resilience: Secure data pipelines; prior to deployment, conduct risk assessments and implement appropriate measures such as least privilege, encryption, logging, and continuous monitoring.

Principle 4. Fairness and Non-Discrimination

• Fairness & Bias Management: Prohibit deployment of known unmitigated bias; require dataset representativeness checks, bias risk registration, and continuous monitoring.
• Prohibited AI Practices: Stakeholder-impacting AI systems shall not be used for manipulative or deceptive techniques, exploitation of vulnerabilities, social scoring or unauthorized biometric surveillance under the EU AI Act Article 5.

Principle 5. Transparency and Explainability

• Transparency & Explainability: Disclose AI interaction, label AI-generated content, and provide risk-appropriate explanations of AI-generated content, including but not limited to: the AI systems’ key functions and purpose, scope and limitations, data sources, and potential risks and related control measures.

Principle 6. Accountability and Governance

• Clear Accountability & Incident Handling: Assign Business and Technical Owners for each AI system; AI governance oversees compliance and risk review; investigate incidents, document root causes and remediation; provide appropriate appeal channels.
• Scope Boundaries, Risk Tiering & Guardrails: Perform risk classification (Low/Medium/High/Prohibited) with governance review; define approved use boundaries, and enforce allowed-use rules and technical guardrails to prevent scope expansion beyond approved boundaries.

Principle 7. Sustainability

• Green AI & Sustainability: Optimize AI system computational efficiency; include sustainability requirements for owned infrastructure, data centers and third-party providers.

Responsible Artificial Intelligence Program

Note: If you have any appeals regarding contesting an AI decision or outcome, please contact the relevant channel “Privacy Protection Procedure” or “Whistleblower.”