iCOMPASS launches KY-AI to help banks manage AI governance

New standards for AI oversight

Financial institutions in Hong Kong are shifting from simple AI experiments to complex tools that handle sensitive data. These systems now frequently recommend specific decisions and act across various connected networks. This shift creates a significant challenge for management regarding how to oversee such autonomous technology.

The Hong Kong Monetary Authority (HKMA) recently introduced the ART Framework to address these concerns. This framework focuses on three core principles: accountability, operational resilience, and trust. Banks must now figure out how to maintain these standards as AI becomes more integrated into their daily operations.

iCOMPASS introduced KY-AI™ at Hong Kong FinTech Week to meet these new expectations. The tool is designed to help banks answer critical questions about where AI is operating and what it is permitted to do. It aims to provide a clear picture of who is responsible for AI actions and whether that behavior remains acceptable over time.

The Know–Approve–Test–Watch cycle

KY-AI™ operates on a structured lifecycle called Know–Approve–Test–Watch. This process allows financial institutions to identify every AI system they use, including those provided by third parties. By understanding these tools first, companies can better assess their potential risks and impact on the business.

Once identified, the system helps establish clear ownership and accountability for each AI tool. It ensures that appropriate approvals are in place before a system goes live. This moves the oversight process away from a one-time check and into a continuous, evidence-based workflow.

The tool also enables institutions to track material changes as AI systems evolve. It is designed to detect emerging risks in real time and provide a way for humans to intervene when necessary. This continuous monitoring is intended to build lasting trust in the outcomes delivered by artificial intelligence.

Applying KYC logic to AI

iCOMPASS is leveraging its experience in the AML/KYC space to build this new governance tool. KYC refers to 'Know Your Customer' rules, which are standard procedures banks use to verify the identity and risk level of their clients. The company is now applying that same level of discipline to artificial intelligence.

The company's existing ULTRA™ platform already handles 6-in-1 AML/KYC workflows, including identity verification and risk assessment. By using a similar approach for AI, iCOMPASS aims to make AI oversight a documented process. This ensures that every decision made by an AI system is traceable and linked to specific evidence.

Raymond Moh, CEO of iCOMPASS, said the tool helps banks build confidence. The goal is to ensure that management truly understands the risks and the specific functions of every AI system in their organization.

Training and regional reach

Beyond the KY-AI™ tool, iCOMPASS is expanding its reach through a new collaboration with Cambridge Advisers Hong Kong. They signed a Memorandum of Understanding to deliver digital compliance training through a platform called UltraLearn™. This platform uses AI to provide training records, assessments, and reminders for staff.

The training content covers several critical areas, including cybersecurity, data protection, and professional conduct. It also includes training for Anti-Money Laundering (AML) and other regulatory requirements. This partnership combines Cambridge Advisers' regulatory content with iCOMPASS's AI-powered learning technology.

iCOMPASS currently serves more than 100 financial institutions across the region. The company holds several security certifications, including ISO/IEC 27001:2022 and a SOC 2 Type II attestation. These credentials are intended to demonstrate that their compliance solutions meet rigorous security and trust standards.

Moving toward automated trust

The launch of KY-AI™ signals a broader trend where regulatory compliance is becoming more automated and integrated. Instead of manual checks, banks are looking for 'Compliance Operating Systems' that can manage multiple tasks at once. This includes regulatory intelligence, policy governance, and risk management in one place.

By connecting traditionally fragmented processes into a unified platform, these tools help institutions reduce manual work. They also turn complex regulatory obligations into demonstrable outcomes that can be shown to auditors. This shift is essential for companies that want to use AI without risking regulatory penalties or operational failures.

As the HKMA's ART Framework becomes a standard, the demand for tools like KY-AI™ is expected to grow. Financial institutions will need to prove they have a handle on their AI's behavior. This move toward evidence-linked workflows ensures that as AI becomes more powerful, it remains under human-led governance.