This mixed-methods study looks into how adult learners in Malaysian higher education use AI large language models (LLMs) for academic purposes. By surveying 103 postgraduate students in business-related fields and conducting in-depth interviews with 10 participants from the original pool, we explored usage patterns, ethical considerations, and cultural factors affecting AI adoption. Results show that AI is widely used (87.4%) for various academic tasks, with a moderate positive link between perceived ease of use and academic dishonesty (R2 = 0.26). The qualitative findings highlight three main themes: contextual ethics, boundary rationalization, and cultural influences on technology acceptance. Malaysian adult learners have sophisticated yet flexible ethical frameworks that challenge traditional views on academic authorship and integrity. These frameworks are influenced by professional norms, collectivist values, and the pressures faced by working professionals pursuing higher education. The study suggests a need to rethink assessment methods and institutional policies to recognize AI as a permanent part of education whilst preserving genuine learning. Instead of focusing primarily on AI detection techniques, institutions should develop responsive approaches that balance technological literacy with academic rigor.