Taishin Bank Collaborates with OneDegree Global on Responsible AI Systems Compliant with Taiwan’s Guidelines for the Use of Artificial Intelligence in the Financial Industry

The first Traditional Chinese large language model in the world developed to comply with global AI governance standards

8 August 2024 – Taishin Bank, one of the largest commercial banks in Taiwan, has partnered with OneDegree Global, a leader in cybersecurity and risk management solutions, for a Proof of Concept. This collaboration aims to validate Taishin Bank’s specialized AI system, “Taishin Brain,” ensuring compliance with the Taiwan Financial Supervisory Commission’s (FSC) Guidelines for the Use of Artificial Intelligence in the Financial Industry. This makes Taishin Bank the first in Taiwan’s financial sector to establish a responsible AI system.

As global AI regulation and compliance become increasingly important – including the recent AI guidelines issued by regulators across Asia, the EU AI Act, and the ISO 42001 standard for AI Management Systems released at the end of 2023 – the importance of compliance and security in AI systems has become more pronounced.

Taishin Bank’s “Taishin Brain” integrates its proprietary data and knowledge base, along with the financial and regulatory compliance dataset, to train and validate its AI model. This is the first Traditional Chinese large language model in the world developed to comply with global AI governance standards. The system leverages financial-specific GPT to process diverse data types such as customer service records, images, and forms, while integrating with internal access control systems to protect privacy and data security.

OneDegree’s cybersecurity unit, Cymetrics, uses advanced AI red team and testing solutions in line with the FSC’s guidelines to verify the fairness, security, and privacy protection of the AI model. Employing the latest AI attack techniques and methods, comprehensive and in-depth testing ensures the AI system’s safety and integrity prior to its launch. Cymetrics also integrates these testing solutions with machine learning operations (MLOps) workflows, enabling testing at every development stage, from model deployment to system go-live, to identify and mitigate inherent risks early in the development process.

Steve Sun, CIO of Taishin Financial Holding, stated, “In the process of AI adoption, Taishin Financial Holding prioritizes information security and privacy, operating under ethical and transparent standards. This ensures compliance with regulatory requirements and, with the support of excellent partners, establishes a responsible AI system. This commitment is central to Taishin’s AI development strategy.”

Stanley Chou, CIO of OneDegree, commented, “We are delighted to collaborate with Taishin Bank, assisting in the AI validation of ‘Taishin Brain,’ and leading the way in establishing Taiwan’s first responsible AI system in the financial sector. As the regulatory environment evolves, Cymetrics will support clients in managing AI risks and regulations, ensuring their AI investments are secure, ethical, and compliant.”