This study examines the use of AI-assisted systems, such as DeepL, for translating German–Italian economic texts. The analysis is based on experiences from the Master’s programme Language for Communication in International Enterprise and Organisation at the University of Modena and Reggio Emilia (2023/24–2024/25), within the German Specialised Communication and Translation course. Initially, students translated a selection of corporate report passages independently. Subsequently, they translated the same passages with DeepL. The analysis reveals discrepancies in terminology accuracy, idiomatic and metaphorical expression use, and pragmatic nuances, comparing student and machine translations. This assertion is supported by literature on specialised translation (Biel/Sosoni 2017: 351–361) and AI-assisted translation (O’Brien/Simard/Specia 2012; Yang et al. 2023). The findings indicate that while machine translation enhances efficiency and produces idiomatic sentence structures, human expertise remains essential for metaphors, stylistic register, and culturally specific terms. The study emphasises the importance of integrating human expertise with AI tools, providing insights relevant to translator training and the incorporation of machine assistance into professional practice.