The emergence of generative AI models such as ChatGPT has transformed translation practices, yet their capacity to handle creative and culturally embedded texts remains questionable. This study investigates ChatGPT’s translation quality in fashion slogans by integrating Molina and Albir’s (2002) translation technique taxonomy with Nababan et al.’s (2012) Translation Quality Assessment (TQA) framework. Employing a descriptive qualitative design complemented by quantitative scoring, the research analyzes 175 English–Indonesian fashion slogans produced by ChatGPT, focusing on three dimensions: accuracy, acceptability, and readability. The evaluation results reveal high accuracy (M = 2.71), indicating strong semantic retention, while acceptability (M = 2.54) and readability (M = 2.62) remain at moderate–high levels, suggesting occasional stylistic and pragmatic incongruence in Indonesian. Genre-based and brand-type analyses further indicate that global slogans tend to be more semantically precise yet less idiomatic compared to local ones. Although ChatGPT successfully preserves literal meaning, it frequently underperforms in maintaining rhetorical, emotional, and cultural resonance. The findings highlight ChatGPT’s potential as a supportive translation tool for marketing discourse but reaffirm the necessity of human post-editing to achieve persuasive equivalence. This research contributes to translation studies by offering an integrated analytical framework that bridges micro-level translation techniques with macro-level quality dimensions, expanding current understanding of AI-mediated translation performance in creative and commercial contexts.