idw – Informationsdienst Wissenschaft
Nachrichten, Termine, Experten
Nachrichten, Termine, Experten
The Vision and Imaging Technologies department of the Fraunhofer Heinrich-Hertz-Institut (HHI) demonstrates in its new research article “High quality deepfakes have a heart!” that high-quality deepfake videos have evolved significantly and can exhibit a detectable heartbeat.
Deepfakes—digitally generated or altered videos, typically created using artificial intelligence—are now widespread and increasingly well-engineered. They enable new creative and technological possibilities but also pose risks, such as disinformation or misuse. Until now, an effective approach to detecting deepfakes has been to analyze subtle physiological signals such as heartbeat patterns, which deepfakes usually did not exhibit.
A recent study has fundamentally changed this assumption. Using remote photoplethysmography (rPPG), a technology originally developed for medical pulse measurement via video, researchers were able to demonstrate that modern deepfakes replicate realistic heartbeats. These heartbeats were extracted from the original videos used to generate the deepfakes.
“As the heart beats, blood flows through blood vessels and into the face,” explains co-author Prof. Peter Eisert. “It’s then distributed over the entire facial area, and there is a small change in skin color that we can pick up in genuine footage.”
The researchers validated their results by comparing the pulse signals detected in the deepfake videos with real heart rate measurements from the original material. These heart rate patterns matched exactly, meaning that these artificial videos do indeed retain the authentic physiological signature of the original subjects.
The team calls for the development of new detection methods that analyze more detailed physiological data, such as localized blood flow patterns in facial regions, instead of relying solely on global heartbeat signals. Additionally, solutions like digital fingerprints—cryptographic markers that verify the authenticity and integrity of digital content—could become essential standards.
The full study was published in the journal Frontiers in Imaging. The authors of the article are Clemens Seibold, Eric L. Wisotzky, Arian Beckmann, Benjamin Kossack, Anna Hilsmann, and Peter Eisert. They work at the Fraunhofer HHI in the Vision and Imaging Technologies department.
The research was funded by the German Federal Ministry of Research, Technology and Space (BMFTR) as part of the FakeID project, as well as by the EU framework program Horizon Europe in the Einstein project.
Prof. Dr.-Ing. Peter Eisert
Head of Vision & Imaging Technologies Department
phone: +49 30 31002-614
peter.eisert@hhi.fraunhofer.de
https://www.frontiersin.org/journals/imaging/articles/10.3389/fimag.2025.1504551...
High-quality deepfake videos have evolved considerably and have a detectable heartbeat.
© AdobeStock/The 2R Artificiality
Merkmale dieser Pressemitteilung:
Journalisten
Informationstechnik, Wirtschaft
überregional
Forschungsergebnisse, Wissenschaftliche Publikationen
Englisch
High-quality deepfake videos have evolved considerably and have a detectable heartbeat.
© AdobeStock/The 2R Artificiality
Sie können Suchbegriffe mit und, oder und / oder nicht verknüpfen, z. B. Philo nicht logie.
Verknüpfungen können Sie mit Klammern voneinander trennen, z. B. (Philo nicht logie) oder (Psycho und logie).
Zusammenhängende Worte werden als Wortgruppe gesucht, wenn Sie sie in Anführungsstriche setzen, z. B. „Bundesrepublik Deutschland“.
Die Erweiterte Suche können Sie auch nutzen, ohne Suchbegriffe einzugeben. Sie orientiert sich dann an den Kriterien, die Sie ausgewählt haben (z. B. nach dem Land oder dem Sachgebiet).
Haben Sie in einer Kategorie kein Kriterium ausgewählt, wird die gesamte Kategorie durchsucht (z.B. alle Sachgebiete oder alle Länder).
