Large language models (LLMs) have demonstrated remarkable capability to generate fluent responses to a wide variety of user queries. However, this has also raised concerns about the potential misuse of such texts in journalism, education, and academia. In this study, we strive to create automated systems that can detect machine-generated texts and pinpoint potential misuse. We first introduce a large-scale benchmark M4, which is a multi-generator, multi-domain, and multi-lingual corpus for machine-generated text detection. Through an extensive empirical study of this dataset, we show that it is challenging for detectors to generalize well on instances from unseen domains or LLMs. In such cases, detectors tend to misclassify machine-generated text as human-written. These results show that the problem is far from solved and that there is a lot of room for improvement. We believe that our dataset will enable future research towards more robust approaches to this pressing societal problem. The dataset is available at https://github.com/mbzuai-nlp/M4

We describe our system for SemEval-2022 Task 3 subtask 2 which on detecting the frames used in a news article in a multi-lingual setup. We propose a multi-lingual approach based on machine translation of the input, followed by an English prediction model. Our system demonstrated good zero-shot transfer capability, achieving micro-F1 scores of 53% for Greek (4th on the leaderboard) and 56.1% for Georgian (3rd on the leaderboard), without any prior training on translated data for these languages. Moreover, our system achieved comparable performance on seven other languages, including German, English, French, Russian, Italian, Polish, and Spanish. Our results demonstrate the feasibility of creating a language-agnostic model for automatic framing detection in online news.