China’s recent nationwide obsession with “lobster raising”—referring to the open-source AI agent software OpenClaw—has quickly spread globally. However, industry observers have raised concerns regarding potential unemployment and security regulation issues in Taiwan. Experts suggest this trend will rewrite the division of labor within Taiwan’s industries. Furthermore, because AI agents must take control of computers and access internal files, utilizing foreign platforms could pose severe risks of corporate data leaks and national security threats.
Jen-Chih Huang, an analyst at the International Economic Research Division of the Chung-Hua Institution for Economic Research (CIER), stated that what these “lobsters” replace first is typically not an entire role, but rather the standardized, digitalized, and trackable tasks within it. Examples include administrative compilation, initial customer service responses, report organization, knowledge retrieval, post-meeting processing, basic programming, and internal workflow integration.
Analyst Jen-Chih Huang cited the 2025 report by the World Economic Forum, pointing out that AI and information processing technologies are among the most transformative technologies of the coming years. Companies will accelerate automation while simultaneously increasing their demand for capabilities related to AI, data, and digital workflows. This implies that the core functions of existing professions will transform accordingly; while some manual tasks will decrease, new demands for human capital will emerge.
Consequently, when mapped against Taiwan’s industrial landscape, Analyst Jen-Chih Huang believes the actual impact may unfold as a “three-tier divergence.” The first tier involves the accelerated restructuring of white-collar procedural work, particularly in office administration, customer service, foundational analysis, marketing material production, and internal IT support.
The second tier is a shift in software development models, transitioning from the traditional “workorder to IT department scheduling” paradigm to one where “business units rapidly create tool prototypes using natural language”—a concept referred to as ambient development and employee-built tools.
The third tier represents the potential amplification of Taiwan’s inherent advantages in ICT and manufacturing. Agentic AI requires not only underlying models but also localized computing hardware, cloud and edge deployment, cybersecurity controls, system integration, enterprise data governance, and vertical domain know-how.
Taiwan is not merely on the receiving end of this disruption; it is well-positioned to be a primary supplier of agentic AI infrastructure and industrial solutions. A more accurate assessment is that these “lobsters” will bifurcate Taiwan’s job market into two categories: those who design, manage, audit, and integrate AI agents; and those constrained by fixed processes who can be easily substituted.
Analyst Jen-Chih Huang contends that AI agents such as OpenClaw will not simply “cause unemployment,” but will instead fundamentally rewrite the logic of labor division in Taiwan. The pressure on low-level, repetitive, and digitalizable white-collar tasks will undoubtedly intensify. Concurrently, new demand will surge for AI implementation consultants, workflow design, data governance, system integration, model evaluation, cybersecurity auditing, and vertical-domain agent development.
For Taiwan, the crucial question is not “whether to raise lobsters,” but rather who possesses the capability to integrate them securely into enterprise workflows—cultivating them into controllable and auditable production tools rather than mere conversational novelties.
Tai-Chin Chiang, Director of The Japan Center Sub Committee at the CIER, also cautioned that because AI agents need to execute computer commands and read internal files, deploying foreign platforms could lead to corporate espionage and national security vulnerabilities. However, current legislation—including the Personal Data Protection Act, the Trade Secrets Act, and the Cyber Security Management Act—lacks clear regulatory frameworks regarding accountability, cross-border data transmission, and algorithmic transparency for AI agents. The legal liability when autonomous AI decisions result in damages also remains to be clarified.
Source: Economic Daily News (March 14, 2026). The “Lobster Raising” AI Craze Ignites Job Anxiety and Security Oversight Concerns in Taiwan. https://money.udn.com/money/amp/story/5603/9380064