AI (Artificial Intelligence) has become an indispensable part of our daily lives. According to Sumsub’s latest survey, AI is gradually evolving from a supportive tool into digital entities capable of autonomous decision-making.
The Sumsub survey reveals that 44% of respondents in Greater China have experienced negative impacts related to AI agents, including unauthorised actions, data leaks, and fraud. The survey also highlights a significant lack of consumer awareness — less than half of the respondents could accurately distinguish AI agents from other AI applications. In Hong Kong, this figure drops to just 45%. This knowledge gap directly increases usage risks and blurs the boundaries of authorisation and responsibility.
At the same time, there is no clear market consensus on accountability mechanisms for AI agents. In Hong Kong, 40% of respondents believe that responsibility for AI-related issues should be shared among multiple parties rather than falling on a single stakeholder. Additionally, nearly 80% of Hong Kong respondents are unaware whether platforms have established usage guidelines for AI agents. Even when restrictions exist, 42% of users admit they would still try to continue using the AI agent or attempt to bypass the limitations.
Sumsub notes that as AI agents become more autonomous, the questions of “Who is responsible for the decisions?” and “How should accountability be assigned when problems arise?” will become central issues. Enterprises emphasise that AI should function as a “co-pilot,” operating under human supervision rather than making critical decisions independently.
Building Digital Trust and Verification Mechanisms is Key
Faced with rising risks, consumers widely agree on the need for stronger protective measures. The survey shows that the three key factors for building trust are: enhanced consumer protection on platforms, introduction of additional verification mechanisms, and improved regulatory oversight. Furthermore, over 70% of respondents believe that requiring human confirmation before an AI agent executes an action can significantly increase their sense of security.
Penny Chai, Vice President, APAC at Sumsub, said that local enterprises’ adoption of AI Agents has grown by approximately 15%. Although this is lower than Taiwan’s 20% and Mainland China’s 37%, it still reflects that companies are gradually experimenting with deployment. “The most important aspect of deploying AI Agents is understanding the permissions granted to the AI and which processes are being automated, so that risks arising from AI Agents can be effectively managed.”
Sumsub has earlier launched an AI agent verification solution under its “Know Your Agent (KYA)” framework. By binding AI behaviour to verified human identities, the solution ensures that only authorised systems can perform operations. This strengthens accountability mechanisms and reduces fraud risks.
Leave a comment