<![CDATA[The increasing use of artificial intelligence (AI) in translation offers opportunities for efficient legal document translation. However, AI-generated translations remain challenging because legal documents require terminological precision and contextual accuracy. This study investigates the accuracy of an Indonesian divorce certificate translated into English using ChatGPT based on Nida and Taber’s concepts of formal and dynamic equivalence. A qualitative descriptive design was employed. Translation units from an Indonesian divorce certificate were categorized into legal, administrative, and socio-cultural terms. Each unit was analyzed in terms of lexical form, legal meaning, and contextual appropriateness. Legal terminology was generally translated accurately through formal equivalence when equivalent legal concepts existed in English. Administrative terminology showed mixed results: established governmental terms were generally appropriate, whereas culture-specific concepts presented accuracy problems. Socio-cultural expressions required dynamic equivalence and posed the greatest challenges because they reflected legal and cultural concepts without direct English equivalents. Overall, ChatGPT performed well in translating standardized legal terminology but showed limitations with culturally embedded and institution-specific expressions. ChatGPT has potential for assisting legal document translation, but culturally specific legal and administrative concepts require human evaluation to ensure semantic precision, contextual appropriateness, and legal reliability. This study demonstrates the application of formal and dynamic equivalence to evaluate ChatGPT-generated legal translations and provides practical insights into its strengths and limitations.]]>
