 |
As regulators issue warnings and guidance on rising risks of fraud, data breaches, and misuse, 82% of respondents in Mainland China said that verifying AI agents would boost their confidence in using them.
SHENZHEN, China, May 18, 2026 /PRNewswire/ — Sumsub, a leading global verification and anti-fraud platform, today released its latest research. The results show that despite ongoing fraud and data security risks, consumers are increasingly authorizing AI agents to make decisions and perform actions. Meanwhile, public awareness has not kept pace with the rapid adoption of AI applications, raising ongoing concerns about trust, accountability, and consumer protection in the digital environment.
In recent months, the adoption of AI agents has accelerated rapidly. Their use is primarily concentrated in familiar applications and services, such as e-commerce platforms, banking and payment software, travel and lifestyle services, and social or instant messaging platforms. In these scenarios, AI agents assist users with answering questions, summarizing information, shopping, planning trips, automating routine tasks, and in some cases, performing actions on behalf of users, such as completing bookings or payments.
Sumsub’s Greater China AI Agent Consumer Trust Survey shows that 44% of respondents reported experiencing at least one risk incident related to AI agents, including erroneous operations (20%), unauthorized charges (16%), personal data breaches (20%), fraudulent behavior (16%), or account intrusions (13%). Overall, the complexity and decision-making autonomy of AI functions are continuously increasing, drawing significant attention from regulators worldwide. In April this year, Chinese authorities jointly released the “Measures for the Ethical Review and Service Management of Artificial Intelligence Technology (Trial),” marking a new phase in regulation that moves from principle-based advocacy to institutionalized and standardized frameworks, further emphasizing accountability, security safeguards, and responsible deployment.
Sumsub’s survey data indicates that among 1,050 respondents from Mainland China, Hong Kong, and Taiwan, 37% of Mainland Chinese respondents reported having already allowed AI agents to directly perform actions on their behalf in the first quarter of 2026 alone. While the use of AI agents continues to rise, only 42% of respondents could clearly distinguish AI agents from other types of AI, less than half. This result highlights a significant awareness gap among the public regarding “the scope of actions AI agents can be authorized to perform” and “which critical steps still require human decision-making and oversight.”
Key Findings:
- Among respondents who self-identified as “early adopters of AI agents,” 59% expressed trust in AI agents operating autonomously without human review;
- 80% of Mainland Chinese respondents said they were either unsure (40%) or unaware (40%) whether platforms had established clear rules for AI agent use;
- When service platforms impose restrictions on AI agent use, 62% of Mainland Chinese respondents said they would still use AI agents beyond those restrictions, such as using them as decision-making references, attempting to bypass restrictions, or switching to other platforms that allow AI agent use;
- When AI agents make mistakes, Mainland Chinese respondents’ views on accountability were divided, with no consensus on whether responsibility should fall on developers, platforms, or users. 35% of respondents were more inclined to believe that multiple parties should share responsibility.
Lai Zhuangwei, Strategy Director for Greater China at Sumsub, stated: “The significant value of AI agents lies in improving efficiency, but in terms of accountability and risk management, they can never replace the human role. They should be seen more as a ‘co-pilot,’ playing a supportive role under clear rules and effective oversight, rather than operating independently without supervision. Currently, technology is evolving at an unprecedented pace. Whether for consumers or businesses, the core need is not higher levels of automation, but clear, explainable, and traceable governance mechanisms—including how decisions are made, who bears responsibility, and how to identify and assign accountability when problems arise. Only with clear boundaries of rights and responsibilities and under safe and controllable conditions can the orderly development of AI agent technology be promoted to serve the real economy.“
AI Agents Accelerate Deployment, “Gatekeepers” Are Crucial
For most consumers, trust in AI agents remains highly correlated with the level of risk involved. Consumers in Greater China are increasingly using AI agents, with over 70% of respondents saying they use them at least once a week. Current usage scenarios are still dominated by low-risk, assistive tasks, such as asking questions or generating content, organizing and summarizing information, and end-to-end planning and research for various activities. However, when it comes to financial matters or other high-risk behaviors, trust in AI agents drops significantly. This discrepancy further reflects consumers’ cautious attitude toward the boundaries of AI agent “autonomy,” especially in scenarios that may entail higher risks or more severe consequences.
Meanwhile, 77% of respondents in Greater China prefer the introduction of a “human confirmation mechanism” before AI agents execute actions, particularly in cases involving payments, sensitive data, and irreversible decisions, further highlighting consumer expectations for “human gatekeeping.” This indicates that as AI agents’ autonomy increases, humans must serve as the ultimate “gatekeepers” to ensure clear rights, responsibilities, and accountability.
Digital Trust Requires Verification Mechanisms and Collaborative Efforts Across Society
Respondents generally believe that the following three factors would significantly boost their confidence in using AI agents: platform-level security guarantees (39%), stricter verification mechanisms (38%), and a more robust regulatory framework (37%). Notably, over 80% of respondents in Mainland China said that verifying AI agents would help enhance their confidence in using them.
Trust is also distributed across different entities. Similar proportions of respondents said they trust AI platform developers (39%), government and regulatory agencies (37%), and independent third-party verification service providers (36%) to ensure the safety and accountability of AI agents. This also indicates that related risks cannot be addressed by a single entity alone and require multi-stakeholder collaboration.
Chen Chongxuan, Head of Government Affairs for Asia Pacific at Sumsub, stated: “Effective governance of AI agents requires close collaboration across all sectors of society. Sumsub is committed to working with governments, regulators, and industry partners to clarify relevant standards and norms, strengthen accountability, and enhance public awareness of responsible AI agent use, ensuring that related systems operate within transparent and enforceable security frameworks.“
Sumsub Partners with YouYueDa to Help Chinese Enterprises Build AI Agent Governance and Compliance Capabilities, Supporting Safe and Orderly “Going Global”
In the current digital economy environment, the risks faced by consumers and businesses do not stem solely from AI agents themselves, but also from long-standing issues in the digitalization process, such as identity fraud, unclear permission boundaries, and insufficient verification. As enterprises accelerate their “going global” efforts, these risks are compounded across different jurisdictions, particularly impacting industries involving overseas user identity verification and compliance needs, such as fintech, cross-border payments and remittances, cross-border e-commerce, gaming going global, and ride-sharing.
Against this backdrop, Sumsub, leveraging its decade-long global experience in verification and anti-fraud, has entered into a strategic partnership with YouYueDa, an Asia-Pacific digital solutions expert, to help enterprises build a comprehensive protection system covering governance, compliance, and anti-fraud while introducing AI agents and automation capabilities, while also considering consumer usage habits and trust expectations to drive stable business development.
While supporting enterprises in accelerating their布局 in key markets such as Southeast Asia, Sumsub also helps Chinese companies steadily expand into emerging markets like Latin America and Africa, assisting them in efficiently addressing diverse compliance requirements and risk challenges across different jurisdictions, and supporting safe and orderly global business expansion.
To help enterprises deploy AI agents more safely and effectively, Sumsub earlier launched an AI agent verification solution based on the innovative Know Your Agent (KYA) framework. This solution helps enterprises verify the identity of AI agents, ensuring that only trusted automated actions can perform tasks on behalf of the enterprise. At the same time, by linking AI agents to verified real user identities and managing AI automated behavior, it effectively addresses AI-driven fraud threats.
For more information on AI agent verification and KYA, please visit: https://sumsub.com/blog/know-your-agent/
Survey Methodology:
To assess consumer acceptance and current usage of AI agents, Sumsub commissioned independent market research firm Blackbox to conduct the Greater China AI Agent Consumer Trust Survey in April 2026. The survey covered 1,050 respondents from Mainland China, Hong Kong, and Taiwan, focusing on analyzing their trust levels, willingness to delegate tasks to AI agents across scenarios of varying complexity and risk, and key factors influencing usage and adoption intentions. The report also explored respondents’ actual usage behaviors, awareness of platform rules and restrictions, and current concerns regarding security, privacy, and accountability, which still affect the further adoption of AI agents to some extent.