The rapid advancement of language translation tools and generative artificial intelligence applications has significantly facilitated the production of academic research while simultaneously introducing new challenges to maintaining academic ethics and detecting plagiarism. This study examines how plagiarism rates vary when academic texts are translated into different languages and evaluates the performance of Turnitin, iThenticate, and Grammarly plagiarism detection software in this context. For this study, ten English-language manuscripts published in the Web of Science and Scopus databases between 2023 and 2024 were selected and translated into Spanish, Portuguese, and French, the most frequently used publication languages after English during the same period. The plagiarism rates of both the original and translated texts were comparatively analyzed using the selected software. The findings reveal that existing plagiarism detection tools are largely ineffective at identifying plagiarism in translated texts, and that Grammarly, in particular, has limited detection capabilities, even for English originals. These results suggest that plagiarism through translation can become virtually undetectable by software, posing significant ethical risks in academic publishing. As an alternative solution, the study proposes a self-assessment form to help researchers to evaluate their own academic work processes in terms of ethical standards. This form introduces the Academic Ethics Negative Degree score. The lower the score, the greater the ethical value of the academic work.