The implementation of internet applications has already crossed the language border. It has, for sure, brought lots of advantages, but to some extent has also introduced some side-effect. One of the negative effects of using these applications is cross-languages plagiarism, which is also known as translated plagiarism.In academic institutions, translated plagiarism can be found in various cases, such as: final project, theses, papers, and so forth. In this paper, a model for web-based early detection system for translated plagiarism is proposed and a prototype is developed. The system works by translating the input document (written in Bahasa Indonesian) into English using Google Translate API components, and then search for documents on the World Wide Web repository which have similar contents to the translated document. If found, the system downloads these documents and then do some preprocessing steps such as: removing punctuations, numbers, stop words, repeated words, lemmatization of words, and the final process is to compare the content of both documents using the modified sentence-based detection algorithm (SBDA). The results show that the proposed method has smaller error rate leading to conclusion that it has better accuracy.