Video Deepfakes Now Pulse With Lifelike Heartbeats

The arms race between deepfake creators and detectors has reached a critical turning point. Researchers have discovered that modern deepfakes can now replicate something previously thought impossible: a realistic human heartbeat.

Scientists at Germany’s Fraunhofer Heinrich-Hertz-Institute and Humboldt University of Berlin found that current deepfake videos inadvertently mimic the subtle skin color fluctuations caused by blood flow in the human face, effectively “inheriting” the pulse from the original source video.

“Here we show for the first time that recent high-quality deepfake videos can feature a realistic heartbeat and minute changes in the color of the face, which makes them much harder to detect,” said Dr. Peter Eisert, a professor at the Humboldt University of Berlin and the study’s corresponding author.

This discovery challenges a major detection method currently used to unmask fake videos. Until now, many experts believed the absence of physiological signals like pulse was a reliable marker for identifying deepfakes.

The research team developed a sophisticated pipeline to extract and analyze heart-related signals from videos through remote photoplethysmography (rPPG), which detects blood volume changes beneath the skin. When they applied this technology to their collection of high-quality deepfakes, they were surprised to find that these fake videos displayed pulse signals nearly identical to those in the original videos.

“Our results show that a realistic heartbeat may be added by an attacker on purpose, but can also be ‘inherited’ inadvertently from the driving genuine video. Small variations in skin tone of the real person get transferred to the deepfake together with facial motion, so that the original pulse is replicated in the fake video,” Dr. Eisert explained.

The study analyzed videos from multiple sources, including a custom dataset with ECG measurements and publicly available videos. In all cases, the deepfakes exhibited heart rates that differed from their source videos by only a few beats per minute—well within the margin of error for this technology.

The findings raise serious concerns about the reliability of current deepfake detection systems. As manipulation techniques advance, distinguishing between genuine and synthetic media becomes increasingly difficult, with potentially devastating consequences for misinformation campaigns and political manipulation.

Fortunately, the researchers identified a potential counter-strategy. While deepfakes can now mimic the global pulse seen across the face, they still fail to accurately reproduce the anatomically correct patterns of blood flow within different facial regions.

“Our experiments have shown that current deepfakes may show a realistic heartbeat, but do not show physiologically realistic variations in blood flow across space and time within the face,” said Dr. Eisert. “We suggest that this weakness of state-of-the-art deepfakes should be exploited by the next generation of deepfake detectors.”

This research highlights the persistent cat-and-mouse game between creation and detection technologies. As deepfakes become more sophisticated, detection methods must evolve accordingly, focusing on increasingly subtle physiological markers that are difficult to replicate.

The study, published in Frontiers in Imaging, serves as both a warning about the growing sophistication of synthetic media and a roadmap for developing more robust verification systems in an era where seeing can no longer be equated with believing.