{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,30]],"date-time":"2026-09-30T06:56:36Z","timestamp":1790751396680,"version":"4.1.0"},"reference-count":43,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,9,30]],"date-time":"2026-09-30T00:00:00Z","timestamp":1790726400000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Artif. Intell."],"abstract":"<jats:p>The rise of deepfake technology has fuelled the need for robust detection methods to counter the increasing sophistication of manipulated multimedia content. Research has shown that humans struggle to identify deepfakes, highlighting the pressing need for AI-based solutions to safeguard the integrity of the information we consume daily. This paper proposes a novel approach CAE-DAD (Convolutional Autoencoder Deepfake Anomaly Detector) which facilitates deepfake detection through visual anomaly identification using convolutional autoencoders (CAEs). Methods like Kernel Density Estimation are used in the latent space to widen the anomaly score gap between real and fake images. The model is trained only on real face images. Existing models like VGG16 which are known for their feature extraction capabilities as incorporated as part of our encoder. The model is benchmarked against state-of-the-art methods and establish a comprehensive set of evaluation metrics, including accuracy, precision, recall, F1 score. The model achieves comparable performance to existing two-class training methods, and achieves 91% accuracy. The proposed anomaly identification using CAEs serves as a valuable addition to the arsenal of tools aimed at mitigating the risks associated with the proliferation of deepfake content.<\/jats:p>","DOI":"10.3389\/frai.2026.1834939","type":"journal-article","created":{"date-parts":[[2026,9,30]],"date-time":"2026-09-30T06:05:25Z","timestamp":1790748325000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Deepfake anomaly detection using kernel density estimation based convolutional autoencoder"],"prefix":"10.3389","volume":"9","author":[{"given":"Manas Ranjan","family":"Prusty","sequence":"first","affiliation":[{"name":"Centre for Cyber Physical Systems, Vellore Institute of Technology","place":["Chennai, India"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Sahil","family":"Amritkar","sequence":"additional","affiliation":[{"name":"School of Computer Science and Engineering, Vellore Institute of Technology","place":["Chennai, India"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ananthakrishnan","family":"Balasundaram","sequence":"additional","affiliation":[{"name":"Centre for Cyber Physical Systems, Vellore Institute of Technology","place":["Chennai, India"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Subhra Rani","family":"Patra","sequence":"additional","affiliation":[{"name":"Department of Information Systems and Operations Management, University of Texas at Arlington","place":["Arlington, TX, United States"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,9,30]]},"reference":[{"key":"ref1","doi-asserted-by":"crossref","first-page":"1245","DOI":"10.1109\/ICESC51422.2021.9532627","volume-title":"2021 Second International Conference on Electronics and Sustainable Communication Systems (ICESC)","author":"Agarwal","year":"2021"},{"key":"ref2","doi-asserted-by":"publisher","first-page":"e1040","DOI":"10.7717\/peerj-cs.1040","article-title":"Deep fake detection using cascaded deep sparse auto-encoder for effective feature selection","volume":"8","author":"Balasubramanian","year":"2022","journal-title":"PeerJ Comput. Sci."},{"key":"ref3","doi-asserted-by":"publisher","first-page":"tyad011","DOI":"10.1093\/cybsec\/tyad011","article-title":"Testing human ability to detect \u2018deepfake\u2019 images of human faces","volume":"9","author":"Bray","year":"2023","journal-title":"J. Cybersecur."},{"key":"ref4","doi-asserted-by":"crossref","first-page":"717","DOI":"10.1007\/978-3-319-45823-6_67","volume-title":"Parallel Problem Solving from Nature \u2013 PPSN XIV","author":"Cao","year":"2016"},{"key":"ref5","doi-asserted-by":"publisher","first-page":"1032","DOI":"10.1109\/TMI.2020.3045295","article-title":"MAMA net: multi-scale attention memory autoencoder network for anomaly detection","volume":"40","author":"Chen","year":"2021","journal-title":"IEEE Trans. Med. Imaging"},{"key":"ref6","doi-asserted-by":"publisher","first-page":"7103","DOI":"10.1002\/int.22582","article-title":"Improved autoencoder for unsupervised anomaly detection","volume":"36","author":"Cheng","year":"2021","journal-title":"Int. J. Intell. Syst."},{"key":"ref7","first-page":"1800","volume-title":"Xception: Deep Learning with Depthwise Separable Convolutions 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)","author":"Chollet","year":"2017"},{"key":"ref8","first-page":"52","volume-title":"\u201cCross-Forgery Analysis of Vision Transformers and CNNs for Deepfake Image Detection,\u201d in Proceedings of the 1st International Workshop on Multimedia AI against Disinformation","author":"Coccomini","year":"2022"},{"key":"ref9","first-page":"5202","volume-title":"Retinaface: single-shot multi-level face Localisation in the wild 2020 IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR)","author":"Deng","year":"2020"},{"key":"ref10","article-title":"The DeepFake detection challenge (DFDC) dataset","author":"Dolhansky","year":"2020"},{"key":"ref11","doi-asserted-by":"publisher","first-page":"104549","DOI":"10.1016\/j.jisa.2026.104549","article-title":"SALT: semantic-guided adaptive latent space truncation sampling watermarking for diffusion models","volume":"101","author":"Guo","year":"2026","journal-title":"J. Inf. Secur. Appl."},{"key":"ref12","first-page":"770","volume-title":"Deep Residual Learning for Image Recognition 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)","author":"He","year":"2016"},{"key":"ref13","doi-asserted-by":"crossref","DOI":"10.1109\/CVPR46437.2021.00434","article-title":"ForgeryNet: a versatile benchmark for comprehensive forgery analysis","author":"He","year":"2021"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1704.04861","article-title":"Mobilenets: efficient convolutional neural networks for mobile vision applications","author":"Howard","year":"2017","journal-title":"arXiv"},{"key":"ref15","doi-asserted-by":"publisher","first-page":"13834","DOI":"10.1109\/TCYB.2021.3127716","article-title":"Self-supervision-augmented deep autoencoder for unsupervised visual anomaly detection","volume":"52","author":"Huang","year":"2022","journal-title":"IEEE Trans. Cybern."},{"key":"ref16","volume-title":"HiDF: A Human-Indistinguishable Deepfake Dataset Zenodo. The 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD '25)","author":"Kang","year":"2025"},{"key":"ref17","first-page":"4217","article-title":"A style-based generator architecture for generative adversarial networks","author":"Karras","year":"2019"},{"key":"ref18","doi-asserted-by":"publisher","first-page":"827","DOI":"10.3390\/agriculture11090827","article-title":"Impacts of background removal on convolutional neural networks for plant disease classification in-situ","volume":"11","author":"Kc","year":"2021","journal-title":"Agriculture"},{"key":"ref19","first-page":"2794","volume-title":"Oc-Fakedect: Classifying Deepfakes using one-class Variational Autoencoder 2020 IEEE\/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)","author":"Khalid","year":"2020"},{"key":"ref20","doi-asserted-by":"publisher","first-page":"93","DOI":"10.3390\/fi13040093","article-title":"iCaps-Dfake: an integrated capsule-based model for deepfake image and video detection","volume":"13","author":"Khalil","year":"2021","journal-title":"Future Internet"},{"key":"ref21","doi-asserted-by":"publisher","first-page":"110713","DOI":"10.1016\/j.bspc.2026.110713","article-title":"P-LDM: partitioned latent diffusion model for memory-efficient 3D medical image translation","volume":"125","author":"Kim","year":"2026","journal-title":"Biomed. Signal Process. Control"},{"key":"ref22","article-title":"Adam: a method for stochastic optimization","author":"Kingma","year":"2014"},{"key":"ref23","doi-asserted-by":"publisher","first-page":"18461","DOI":"10.1007\/s11042-020-10420-8","article-title":"Detecting DeepFake, FaceSwap and Face2Face facial forgeries using frequency CNN","volume":"80","author":"Kohli","year":"2021","journal-title":"Multimed. Tools Appl."},{"key":"ref24","doi-asserted-by":"publisher","first-page":"100054","DOI":"10.1016\/j.jjimei.2021.100054","article-title":"Exposing deepfakes using a deep multilayer perceptron \u2013 convolutional neural network model","volume":"2","author":"Kolagati","year":"2022","journal-title":"Int. J. Inf. Manag. Data Insights"},{"key":"ref25","doi-asserted-by":"publisher","first-page":"100234","DOI":"10.1016\/j.metrad.2026.100234","article-title":"Cycle-LVDM: cycle-consistent latent variational diffusion model for unpaired medical image translation","author":"Kui","year":"2026","journal-title":"Meta-Radiology"},{"key":"ref26","article-title":"ADD: frequency attention and multi-view based knowledge distillation to detect low-quality compressed deepfake images","author":"Le","year":"2021"},{"key":"ref27","first-page":"3204","article-title":"Celeb-DF: a large-scale challenging dataset for deepfake forensics","author":"Li","year":"2020"},{"key":"ref28","first-page":"3730","article-title":"Deep learning face attributes in the wild","author":"Liu","year":"2014"},{"key":"ref29","first-page":"1","volume-title":"BMVC 2015 - Proceedings of the British Machine Vision Conference 2015","author":"Parkhi","year":"2015"},{"key":"ref30","doi-asserted-by":"publisher","first-page":"84","DOI":"10.11591\/ijai.v10.i1.pp84-92","article-title":"Effect of data-augmentation on fine-tuned CNN model performance","volume":"10","author":"Poojary","year":"2021","journal-title":"IAES Int. J. Artif. Intell. IJ-AI"},{"key":"ref31","doi-asserted-by":"crossref","first-page":"271","DOI":"10.1109\/IBCAST51254.2021.9393234","volume-title":"2021 International Bhurban Conference on Applied Sciences and Technologies (IBCAST)","author":"Qurat-ul-ain","year":"2021"},{"key":"ref32","doi-asserted-by":"publisher","first-page":"7422","DOI":"10.1038\/s41598-023-34629-3","article-title":"Deep fake detection and classification using error-level analysis and deep learning","volume":"13","author":"Rafique","year":"2023","journal-title":"Sci. Rep."},{"key":"ref33","doi-asserted-by":"publisher","first-page":"25494","DOI":"10.1109\/ACCESS.2022.3154404","article-title":"Deepfake detection: a systematic literature review","volume":"10","author":"Rana","year":"2022","journal-title":"IEEE Access"},{"key":"ref34","first-page":"1","volume-title":"Attentive Autoencoders For Improving Visual Anomaly Detection 2021 IEEE International Conference on Autonomous Systems (ICAS)","author":"Ravi","year":"2021"},{"key":"ref35","doi-asserted-by":"publisher","first-page":"9820","DOI":"10.3390\/app12199820","article-title":"A novel deep learning approach for deepfake image detection","volume":"12","author":"Raza","year":"2022","journal-title":"Appl. Sci."},{"key":"ref36","year":"2019"},{"key":"ref37","first-page":"10674","volume-title":"High-Resolution Image Synthesis with Latent Diffusion Models 2022 IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR)","author":"Rombach","year":"2022"},{"key":"ref38","first-page":"1","article-title":"FaceForensics++: learning to detect manipulated facial images","author":"R\u00f6ssler","year":"2019"},{"key":"ref39","doi-asserted-by":"publisher","first-page":"1","DOI":"10.14569\/IJACSA.2021.0120717","article-title":"Anomaly detection on medical images using autoencoder and convolutional neural network","volume":"12","author":"Siddalingappa","year":"2021","journal-title":"Int. J. Adv. Comput. Sci. Appl."},{"key":"ref40","article-title":"Very deep convolutional networks for large-scale image recognition","author":"Simonyan","year":"2014"},{"key":"ref41","doi-asserted-by":"publisher","first-page":"74","DOI":"10.1016\/j.gltp.2022.04.017","article-title":"Deepfake detection in digital media forensics","volume":"3","author":"Vamsi","year":"2022","journal-title":"Glob. Transit. Proc."},{"key":"ref42","doi-asserted-by":"publisher","first-page":"599","DOI":"10.14569\/IJACSA.2021.0120670","article-title":"Classification Model Evaluation Metrics","volume":"12","author":"Vujovic","year":"2021","journal-title":"Int. J. Adv. Comput. Sci. Appl."},{"key":"ref43","first-page":"1869","volume-title":"A Two-Stage Autoencoder For Visual Anomaly Detection 2021 IEEE International Conference on Image Processing (ICIP)","author":"Zhu","year":"2021"}],"container-title":["Frontiers in Artificial Intelligence"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/frai.2026.1834939\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,30]],"date-time":"2026-09-30T06:05:29Z","timestamp":1790748329000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/frai.2026.1834939\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9,30]]},"references-count":43,"alternative-id":["10.3389\/frai.2026.1834939"],"URL":"https:\/\/doi.org\/10.3389\/frai.2026.1834939","relation":{},"ISSN":["2624-8212"],"issn-type":[{"value":"2624-8212","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,9,30]]},"article-number":"1834939"}}