Another factor surpassed by AI to produce Deepfake

They create a tool that imitates the heartbeat.

If you are reading this post, it means that you do not live in a cave and you must have already heard about Deepfakes, those fake videos that use artificial intelligence to simulate someone's face and voice with terrifying precision, but I would tell you that now they go even further and even manage to simulate the beats of a human heart.

Yes, I am talking about Deep fakes that fool biometric sensors by reproducing the same signals that the human body emits and this was recently a promising way to detect fake videos, but it could have already become obsolete.

For years, researchers used a technique called RPP to identify Deep Fixes. It detects subtle color alterations in facial skin caused by heartbeats. As deepfakes are generated frame by frame, they did not usually contain this natural blood rhythm, which gave them away, but researchers at Humboldt University in Berlin showed that this is changing. Using advanced face swapping tools, they managed to create Deepfakes so well made that they display apparently real heartbeats.

New defenses, new attacks.

And most impressively, those signals were not added manually, they came automatically along with the original face used in the fake, the study revealed two chilling paths, artificial addition of heartbeats, where the creator of the deepfake simulates color variations manually, as if it were a human pulse; Inheritance from the original video, where the real variations of the original face are copied along with the facial movement and the heart of the real person continues to beat in the face of another, the result an RPP-based detector developed by the scientists was completely fooled into reporting accurate heartbeats where there was no real human being.

Despite the scare, the researchers did not give up, they realized that although the general heart rate is being well falsified, the distribution of blood flow through the face is still imperfect, in a real human being the blood circulates in a subtly different way through the cheeks, forehead and chin and that complexity still defies algorithms, that is why the next generation of detectors must go beyond the single heartbeat and begin to map the spatial pattern of the pulse, region by region as if it were a heat map of moving blood.

The new generation of deepfakes is not just an internet joke, it has already been used for extortion, political manipulation and the spread of disinformation, now with fake heartbeats and built-in biometric signals this technology enters a new phase, it can mock even the sensors that were based on the human body to protect against fraud. Security systems, digital banks, video interviews, even criminal investigations are at risk of being deceived by an image that literally seems alive.