Governing AI Across Markets: How Client-Origin Expectations Shape Human-AI Collaboration in Globally Oriented Advertising Agencies in Asia

Global advertising agencies now deploy AI systems that make millions of campaign decisions daily, yet their clients hold fundamentally different expectations about the acceptable level of algorithmic authority. A European financial services client under GDPR and a performance-oriented Southeast Asian brand illustrate this divergence, which reshapes how human-AI collaboration must be governed in practice. Drawing on multi-sited ethnographic fieldwork across three globally oriented advertising agencies in Asia, comprising participant observation and semi-structured interviews with 30 practitioners, this paper develops evidence that AI governance is not standardized but relational: it is shaped by client-origin governance expectations as attributed by agency-side practitioners. We develop a three-dimensional framework covering who decides (delegation architecture), how much they are trusted (trust calibration), and who answers for outcomes (accountability navigation), and introduce client-origin governance expectations as the moderating construct explaining this variation across clients, one that capability-based and institutional accounts alone cannot address. The paper specifies a boundary condition in Human-AI Complementarity Theory and the Hybrid Intelligence Framework: neither fully accounts for the external relational conditions through which governance is constituted. All claims about client expectations reflect agency-side practitioner attributions, not verified client positions. Practical implications are developed for global agencies, platform developers, clients, and brand managers.