
SEOUL / BEIJING — As autonomous artificial intelligence agents transition from conversational chatbots into software capable of booking tickets, making phone calls, and purchasing goods, East Asia’s two major tech powers are taking starkly different approaches to governing them. While Beijing has spent the past six months constructing a state-backed technical and regulatory architecture for on-device AI, South Korea’s nascent agent market is increasingly bogged down by security workarounds, platform blocks, and a lack of legal guardrails.
Beijing’s Regulatory Blueprint: Identity, Payments, and App Permissions
China’s push stems from a national industrial target set by the State Council to achieve over 70% adoption of AI-enabled smart terminals and agents by 2027, rising above 90% by 2030. With market researcher IDC projecting that AI smartphones will make up 53% of China’s handset shipments this year—roughly 147 million devices—regulators have moved quickly to treat on-device agents as core telecommunications infrastructure rather than standalone applications.
In late October, China’s Ministry of Industry and Information Technology (MIIT) brought embedded smartphone agents under its mandatory network access licensing regime. The rule effectively enforces technical standards issued earlier in June by the State Administration for Market Regulation, requiring AI agents to carry standardized digital identity codes before devices can be sold commercially. More than 2,000 key industrial agents have already been registered under the framework.
Financial regulators have paired identity verification with payment rules. Under guidance from the People’s Bank of China, the Payment & Clearing Association of China introduced a "Know Your Agent" (KYA) framework in August. The policy outlines a phased roadmap moving from human-approved transactions—such as Alipay’s AI payment feature, which surpassed 300 million transactions by mid-year—toward bounded autonomous payments within pre-set user spending limits, while keeping legal and financial liability firmly attached to the human account holder and payment provider.
To prevent unauthorized screen-scraping and fraud, regulators also barred agents from simulating clicks inside third-party apps without explicit digital signature consent from app developers. That restriction directly affected devices like ByteDance’s Doubao smartphone, which was blocked from automating tasks inside Tencent’s WeChat and Ant Group’s Alipay. At the same time, antitrust rules prohibit dominant payment platforms from locking out rival AI agents, forcing super-apps and hardware makers such as Huawei, Xiaomi, and Honor into standardized API partnerships with tech giants including Tencent and Alibaba.
South Korea’s Patchwork Workarounds and Platform Friction
In contrast, South Korea’s consumer agent sector has expanded rapidly in user demand without matching institutional infrastructure. Applications such as local reservation assistant Pickle, xAI’s Grok Bot, OpenAI’s Dots, and Hashed-backed platform vooy have gained strong traction among Korean users, with Dots ranking third globally in weekly active users.
Yet without standardized agent authentication or delegated payment protocols, startups have resorted to collecting users' raw login credentials for major portals, e-commerce sites, and government services. Cybersecurity specialists warn that centralizing plain-text passwords and personal histories creates severe breach risks alongside emerging threats like prompt-injection attacks.
Incumbent platforms have responded defensively. Rail operator KORAIL recently moved to block automated AI ticketing bots, while Kakao halted an unauthorized KakaoTalk integration by Pickle within a day of its launch to review privacy and legal compliance. Meanwhile, both Naver and Kakao are preparing to roll out their own proprietary AI assistants, deepening friction between independent startups and established ecosystem gatekeepers.
Regulatory gaps are already producing real-world missteps. Pickle recently apologized after placing automated phone calls without explicit user sign-off or disclosing its AI identity to recipients, while domestic payment trials—such as Shinhan Card and Mastercard’s authenticated agentic commerce pilot in March—remain limited in scope without formal liability rules.
Industry observers note that establishing baseline standards—such as verifiable agent credentials, mandatory AI disclosure on voice calls, explicit user confirmation for financial transactions, and clear rules governing third-party API access—will be critical if South Korea hopes to turn early consumer enthusiasm into a sustainable market rather than a series of regulatory and security disputes.
[Copyright (c) Global Economic Times. All Rights Reserved.]




























