Deepfakes evolve again: now they can fake human heartbeat and escape detection tools

Researchers in Berlin reveal that the latest generation of deepfake videos now successfully mimics a realistic heartbeat, making recognition significantly more difficult.

Deepfake creators use deep learning to manipulate videos and audio files. They alter facial expressions and gestures, for example swapping these between different people. Their purpose isn’t necessarily malign: for example, apps that can turn you into a cat or digitally age you are immensely popular and harmless fun, but still there are people our there that use it in a malicious way.

Algorithms for recognising fake videos have until recently relied on remote pulse reading to distinguish them from genuine videos of people through so-called “remote photoplethysmography” (rPPP), a technique that detects subtle changes in skin colour associated with blood circulation. But new research published in the journal Frontiers in Imaging shows that this technology is no longer reliable and outdated. This worrying development makes it necessary for deepfake detectors to become more powerful.

“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 Prof. Peter Eisert from Humboldt University of Berlin, the lead author of the study.

“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,” said Eisert. This revelation poses new challenges for the fight against misinformation, especially in the context of political campaigns and online manipulation.

However, the researchers are optimistic: while the pulse as a whole can be faked, local changes in facial blood flow remain difficult to mimic. These are what the next generations of deepfake protection algorithms should focus on.

“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 Eisert. “We suggest that this weakness of state-of-the-art deepfakes should be exploited by the next generation of deep fake detectors.”