{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,4]],"date-time":"2026-06-04T07:02:29Z","timestamp":1780556549021,"version":"3.54.1"},"reference-count":19,"publisher":"Springer Fachmedien Wiesbaden GmbH","issue":"2","license":[{"start":{"date-parts":[[2026,2,20]],"date-time":"2026-02-20T00:00:00Z","timestamp":1771545600000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/deed.de"},{"start":{"date-parts":[[2026,2,20]],"date-time":"2026-02-20T00:00:00Z","timestamp":1771545600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/deed.de"}],"funder":[{"name":"Hochschule Heilbronn"}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["HMD"],"published-print":{"date-parts":[[2026,4]]},"abstract":"<jats:title>Zusammenfassung<\/jats:title>\n                  <jats:p>Deepfakes, KI-generierte synthetische Medien, die Menschen realistisch manipulieren, stellen ein wachsendes Sicherheitsrisiko f\u00fcr Unternehmen dar. Ein Beispiel hierf\u00fcr ist der sogenannte CEO-Fraud. In einem bekannt gewordenen Fall wurde ein Manager von einem vermeintlichen CEO zu einer Millionen\u00fcberweisung bewegt, was einen finanziellen Verlust von \u00fcber 35\u00a0Mio. USD zur Folge hatte. Eine schnelle und zuverl\u00e4ssige Erkennung von Deepfakes wird f\u00fcr Unternehmen daher immer bedeutender. In diesem Beitrag wird ein Ansatz vorgestellt, der auf Vision-Transformer und einem expertenbasierten Ensemble-Modell aufbaut, das mit Hilfe von Few-Shot-Learning schnell auf neue Generationsmodelle reagieren kann, ohne ein vollst\u00e4ndiges Retraining durchf\u00fchren zu m\u00fcssen. Unternehmen erm\u00f6glicht dieser Ansatz eine ressourcenschonende, skalierbare und anpassungsf\u00e4hige Deepfake-Erkennung. So lassen sich visuelle Manipulationen bei unbekannten Angriffsmustern fr\u00fchzeitig erkennen und Risiken in der digitalen Kommunikation wirksam reduzieren. Der entwickelte Prototyp, ein Ensemble-Modell, das auf einer Mehrheitsentscheidung der Experten beruht, wurde an ausgew\u00e4hlten Deepfake-Datens\u00e4tzen evaluiert und erzielte eine Gesamtgenauigkeit, die im Vergleich zur Baseline um 57,6\u202f% gestiegen ist.<\/jats:p>","DOI":"10.1365\/s40702-026-01256-1","type":"journal-article","created":{"date-parts":[[2026,2,20]],"date-time":"2026-02-20T16:38:18Z","timestamp":1771605498000},"page":"456-471","update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["Deepfake-Erkennung auch ohne gro\u00dfe Datens\u00e4tze: Ein Prototyp f\u00fcr Organisationen","Deepfake Detection Without Large Datasets: A Prototype for Organizations"],"prefix":"10.1365","volume":"63","author":[{"given":"Jan","family":"Czemmel","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Faisal","family":"Karim","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-2509-7422","authenticated-orcid":false,"given":"Jochen","family":"G\u00fcnther","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-9319-1437","authenticated-orcid":false,"given":"Carsten","family":"Lanquillon","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"93","published-online":{"date-parts":[[2026,2,20]]},"reference":[{"key":"1256_CR1","doi-asserted-by":"publisher","DOI":"10.34740\/KAGGLE\/DSV\/12366330","volume-title":"FaceForencis++ extracted frames [Data set]. Kaggle","author":"7elmy A","year":"2025","unstructured":"7elmy A (2025) FaceForencis++ extracted frames [Data set]. Kaggle https:\/\/doi.org\/10.34740\/KAGGLE\/DSV\/12366330"},{"key":"1256_CR2","doi-asserted-by":"publisher","first-page":"1400024","DOI":"10.3389\/fdata.2024.1400024","volume":"7","author":"E Altuncu","year":"2024","unstructured":"Altuncu\u00a0E, Franqueira\u00a0VNL, Li\u00a0S (2024) Deepfake: definitions, performance metrics and standards, datasets, and a\u00a0meta-review. Front Big Data 7:1400024. https:\/\/doi.org\/10.3389\/fdata.2024.1400024","journal-title":"Front Big Data"},{"issue":"12","key":"1256_CR3","doi-asserted-by":"publisher","first-page":"6819","DOI":"10.1177\/14614448241253138","volume":"27","author":"A Birrer","year":"2024","unstructured":"Birrer\u00a0A, Just\u00a0N (2024) What we know and don\u2019t know about deepfakes: an investigation into the state of the research and regulatory landscape. New Media Soc 27(12):6819\u20136838. https:\/\/doi.org\/10.1177\/14614448241253138","journal-title":"New Media Soc"},{"key":"1256_CR4","unstructured":"Bitkom (2024) Angriffe auf die deutsche Wirtschaft nehmen zu. https:\/\/www.bitkom.org\/Presse\/Presseinformation\/Angriffe-auf-die-deutsche-Wirtschaft-nehmen-zu. Zugegriffen: 30. Okt. 2025"},{"key":"1256_CR5","unstructured":"Bundeskriminalamt (2024) Number of cases of cybercrime recorded by police in Germany from 2007 to 2023 [Graph]. Statista. https:\/\/www.statista.com\/statistics\/1360141\/cyber-crime-cases-recorded-police-germany\/. Zugegriffen: 30. Okt. 2025"},{"key":"1256_CR6","doi-asserted-by":"publisher","first-page":"907199","DOI":"10.3389\/fsoc.2022.907199","volume":"7","author":"A Eberl","year":"2022","unstructured":"Eberl\u00a0A, K\u00fchn\u00a0J, Wolbring\u00a0T (2022) Using deepfakes for experiments in the social sciences - A pilot study. Front Sociol 7:907199. https:\/\/doi.org\/10.3389\/fsoc.2022.907199","journal-title":"Front Sociol"},{"issue":"3","key":"1256_CR7","doi-asserted-by":"publisher","first-page":"585","DOI":"10.3390\/electronics13030585","volume":"13","author":"LY Gong","year":"2024","unstructured":"Gong\u00a0LY, Li\u00a0XJ (2024) A contemporary survey on deepfake detection: Datasets, algorithms, and challenges. Electronics 13(3):585. https:\/\/doi.org\/10.3390\/electronics13030585","journal-title":"Electronics"},{"key":"1256_CR8","unstructured":"Iakubovskyi D (2025) deepfake_vs_real_image_detection. Hugging Face. https:\/\/huggingface.co\/dima806\/deepfake_vs_real_image_detection. Zugegriffen: 8. Okt. 2025"},{"key":"1256_CR9","doi-asserted-by":"publisher","first-page":"5","DOI":"10.1016\/j.patcog.2007.10.015","volume":"41","author":"HRA Ko","year":"2008","unstructured":"Ko\u00a0HRA, Sabourin\u00a0R, Souza Britto\u00a0AS (2008) From dynamic classifier selection to dynamic ensemble selection. Pattern Recognit 41:5. https:\/\/doi.org\/10.1016\/j.patcog.2007.10.015","journal-title":"Pattern Recognit"},{"key":"1256_CR10","doi-asserted-by":"publisher","first-page":"83","DOI":"10.1109\/WACVW.2019.00020","volume-title":"Exploiting visual artifacts to expose deepfakes and face manipulations","author":"F Matern","year":"2019","unstructured":"Matern\u00a0F, Riess\u00a0C, Stamminger\u00a0M (2019) Exploiting visual artifacts to expose deepfakes and face manipulations. IEEE Winter Applications of Computer Vision Workshops (WACVW), S\u00a083\u201392 https:\/\/doi.org\/10.1109\/WACVW.2019.00020"},{"key":"1256_CR11","doi-asserted-by":"publisher","first-page":"45","DOI":"10.2753\/MIS0742-1222240302","volume":"24","author":"K Peffers","year":"2007","unstructured":"Peffers\u00a0K, Tuunanen\u00a0T, Rothenberger\u00a0M, Chatterjee\u00a0S (2007) A design science research methodology for information systems research. J\u00a0Manag Inf Syst 24:45\u201377","journal-title":"J Manag Inf Syst"},{"key":"1256_CR12","doi-asserted-by":"publisher","unstructured":"Pei\u00a0G et\u00a0al. (2024) Deepfake Generation and Detection: A Benchmark and Survey. arXiv preprint. https:\/\/doi.org\/10.48550\/arXiv.2403.17881","DOI":"10.48550\/arXiv.2403.17881"},{"key":"1256_CR13","doi-asserted-by":"publisher","unstructured":"Radford A et al. (2021) Learning Transferable Visual Models From Natural Language Supervision. arXiv preprint. https:\/\/doi.org\/10.48550\/arXiv.2103.00020","DOI":"10.48550\/arXiv.2103.00020"},{"issue":"4","key":"1256_CR14","doi-asserted-by":"publisher","first-page":"1066","DOI":"10.1108\/JFC-04-2022-0090","volume":"30","author":"A de Rancourt-Raymond","year":"2023","unstructured":"de Rancourt-Raymond\u00a0A, Smaili\u00a0N (2023) The unethical use of deepfakes. JFC 30(4):1066\u20131077. https:\/\/doi.org\/10.1108\/JFC-04-2022-0090","journal-title":"JFC"},{"key":"1256_CR15","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1109\/ICCV.2019.00009","volume-title":"FaceForensics++: learning to detect manipulated facial images","author":"A R\u00f6ssler","year":"2019","unstructured":"R\u00f6ssler\u00a0A, Cozzolino\u00a0D, Verdoliva\u00a0L, Riess\u00a0C, Thies\u00a0J, Nie\u00dfner\u00a0M (2019) FaceForensics++: learning to detect manipulated facial images. IEEE\/CVF International Conference on Computer Vision (ICCV), S\u00a01\u201311 https:\/\/doi.org\/10.1109\/ICCV.2019.00009"},{"key":"1256_CR16","unstructured":"Sakthi P (2025) Deep-Fake-Detector-v2-Model. Hugging Face. https:\/\/huggingface.co\/prithivMLmods\/Deep-Fake-Detector-v2-Model. Zugegriffen: 8. Okt. 2025"},{"key":"1256_CR17","volume-title":"Cybercrime-Trends 2025","author":"SoSafe","year":"2025","unstructured":"SoSafe (2025a) Cybercrime-Trends 2025. SoSafe"},{"key":"1256_CR18","volume-title":"Human Risk Review 2024\u20132025: Die Cyber-Bedrohungslage: Experteneinblicke und Strategien","author":"SoSafe","year":"2025","unstructured":"SoSafe (2025b) Human Risk Review 2024\u20132025: Die Cyber-Bedrohungslage: Experteneinblicke und Strategien. SoSafe"},{"key":"1256_CR19","doi-asserted-by":"publisher","first-page":"102784","DOI":"10.1016\/j.technovation.2023.102784","volume":"125","author":"L Whittaker","year":"2023","unstructured":"Whittaker\u00a0L, Mulcahy\u00a0R, Letheren\u00a0K, Kietzmann\u00a0J, Russell-Bennett\u00a0R (2023) Mapping the deepfake landscape for innovation: A\u00a0multidisciplinary systematic review and future research agenda. Technovation 125:102784. https:\/\/doi.org\/10.1016\/j.technovation.2023.102784","journal-title":"Technovation"}],"container-title":["HMD Praxis der Wirtschaftsinformatik"],"original-title":[],"language":"de","link":[{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1365\/s40702-026-01256-1.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/article\/10.1365\/s40702-026-01256-1","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/link.springer.com\/content\/pdf\/10.1365\/s40702-026-01256-1.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,6,4]],"date-time":"2026-06-04T06:25:20Z","timestamp":1780554320000},"score":1,"resource":{"primary":{"URL":"https:\/\/link.springer.com\/10.1365\/s40702-026-01256-1"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,2,20]]},"references-count":19,"journal-issue":{"issue":"2","published-print":{"date-parts":[[2026,4]]}},"alternative-id":["1256"],"URL":"https:\/\/doi.org\/10.1365\/s40702-026-01256-1","relation":{},"ISSN":["1436-3011","2198-2775"],"issn-type":[{"value":"1436-3011","type":"print"},{"value":"2198-2775","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,2,20]]},"assertion":[{"value":"2 November 2025","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"5 February 2026","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"20 February 2026","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"J.\u00a0Czemmel, F.\u00a0Karim, J.\u00a0G\u00fcnther und C.\u00a0Lanquillon geben an, dass kein Interessenkonflikt besteht.","order":1,"name":"Ethics","group":{"name":"EthicsHeading","label":"Interessenkonflikt"}}]}}