Sumsub survey reveals 94% of Hong Kong businesses use AI agents, but only 36% can provide decision-making evidence

Sumsub recently released the "Sumsub Digital Trust Status in the Asia-Pacific Region: AI Governance Benchmark Report," which indicates that 94% of Hong Kong enterprises have implemented or are piloting multi-step AI systems. However, only 36% of these enterprises can provide reliable and tamper-proof records to explain the decision-making processes of AI, highlighting significant challenges in AI governance.

Current Status and Challenges of AI Applications

The report points out that as the capabilities of AI agents continue to improve, enterprises are accelerating the application of AI decision-making without direct human oversight. However, this has also led to a lack of traceability in the decision-making process, increasing risks related to compliance, operations, and customer trust. The "accountability asymmetry issue" mentioned in the study reflects the inability of enterprises to take responsibility for AI decisions, failing to clearly explain to regulators or customers how AI makes decisions.

Main Findings of the Report

According to the survey, the performance of Hong Kong enterprises in AI governance exhibits several notable characteristics compared to other countries in the Asia-Pacific region:

• Expansion of AI Applications: 95% of Hong Kong enterprises reported that they expanded the scale of AI applications or enhanced their autonomy in the past year, slightly above the regional average (92%).
• Shared Accountability Model: 40% of Hong Kong enterprises indicated that the responsibility for AI behavior is shared among multiple business departments, higher than the regional average (27%).
• Cautious Application in High-Risk Workflows: 89% of Hong Kong enterprises are willing to delegate low-risk tasks to AI, but remain cautious when it comes to financial decision-making.
• Need for Improved Governance Maturity: The overall AI governance score for Hong Kong enterprises is 66.7, slightly lower than the regional average score (67.1).

Penny Chai, Vice President of Sumsub Asia-Pacific, stated that while Hong Kong enterprises are not lacking direction in their strategies to promote AI applications, they still need to carefully consider the traceability mechanisms of decisions when expanding applications to mitigate potential operational risks.

Future Outlook and Recommendations

As enterprises increasingly rely on AI applications, establishing a robust traceability and security framework becomes a prerequisite for the safe use of AI in high-risk decision-making. The report recommends that enterprises drive structural transformation by applying a unified trust infrastructure, ensuring that every automated action can be traced back to authorized AI agents and responsible supervisory personnel.

For more details and to download the "Sumsub Digital Trust Status in the Asia-Pacific Region: AI Governance Benchmark Report," please visit the Sumsub official website.