Deepfake social media texts (GPT-2 samples included) can already be found, though there is still no misuse episode on them.ĭeepfake detecting strategies are continuously developed, from deepfake video to audio and text detection methods. This is more than ever necessary for the text generation field: in 2019, for the first time, a generative model (GPT-2 language model ) showed incredible text generation capabilities that deeply worries the research community: and proved that humans seem unable to identify automatically generated text (their accuracy is near random guessing, i.e. Therefore, it is necessary to continuously probe the generative model’s ability to produce deceiving multimedia by enhancing and developing appropriate detectors. The generation and sharing of deepfake multimedia over social media, tricking people into believing that they are human-generated, have already caused distress in several fields (such as politics and science). The term deepfake is a portmanteau of deep learning and fake it refers to AI-generated multimedia (images, videos, audios, and texts) that are potentially deceptive, although good usages of deepfakes can be found. įunding: This work was supported in part by the EU H2020 Program under the scheme INFRAIA-01-2018-2019 Research and Innovation action grant agreement #871042 SoBigData++ European Integrated Infrastructure for Social Mining and Big Data Analytics and AI4Media - A European Excellence Centre for Media, Society and Democracy, H2020 ICT-48-2020, grant 95191.Ĭompeting interests: The authors have declared that no competing interests exist.ĭuring the last decade, the social media platforms-developed to connect people and make them share their ideas and opinions through multimedia contents (like images, video, audio, and texts)-have also been used to manipulate and alter the public opinion thanks to bots, i.e., computer programs that control a fake social media account as a legitimate human user would do: by “liking”, sharing and posting old or new media which could be real, forged through simple techniques (e.g., editing of a video, use of gap-filling texts and search-and-replace methods) or deepfake.ĭeep learning is a family of machine-learning methods that use artificial neural networks to learn a hierarchy of representations, from low to high non-linear features representation, of the input data. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.ĭata Availability: The data underlying the results presented in the study are available from Kaggle at. Received: AugAccepted: ApPublished: May 13, 2021Ĭopyright: © 2021 Fagni et al. Sam Houston State University, UNITED STATES We hope that TweepFake can offer the opportunity to tackle the deepfake detection on social media messages as well.Ĭitation: Fagni T, Falchi F, Gambini M, Martella A, Tesconi M (2021) TweepFake: About detecting deepfake tweets. Lastly, we evaluated 13 deepfake text detection methods (based on various state-of-the-art approaches) to both demonstrate the challenges that Tweepfake poses and create a solid baseline of detection techniques. The dataset is publicly available on Kaggle. We also randomly selected tweets from the humans imitated by the bots to have an overall balanced dataset of 25,572 tweets (half human and half bots generated). The bots are based on various generation techniques, i.e., Markov Chains, RNN, RNN+Markov, LSTM, GPT-2. We collected tweets from a total of 23 bots, imitating 17 human accounts. It is real in the sense that each deepfake tweet was actually posted on Twitter. With the aim of helping the research in this detection field, we collected the first dataset of real deepfake tweets, TweepFake. However, to the best of our knowledge no one has ever addressed the detection of machine-generated texts on social networks like Twitter or Facebook. To prevent this, it is crucial to develop deepfake social media messages detection systems.

Adversaries can exploit these tremendous generative capabilities to enhance social bots that will have the ability to write plausible deepfake messages, hoping to contaminate public debate. Since then, ever more powerful text generative models have been developed. The recent advances in language modeling significantly improved the generative capabilities of deep neural models: in 2019 OpenAI released GPT-2, a pre-trained language model that can autonomously generate coherent, non-trivial and human-like text samples.

Deep sentinel and twitter