Heartbeats!

May 3rd, 2025| Topic: RaMbLeS | 0

For a long time we thought heartbeats would save our identities. How, you ask?

Cybersecurity experts thought they had deepfake videos all figured out. You see, there are subtle color changes to our skins caused by the pulsation of blood through our vascular system, timed with our heartbeats. But AI wasn’t explicitly programmed to mimic these nuanced changes caused by blood flow, so fake videos would lack these physiological signals.

Deepfakes have become ubiquitous in our modern society, with both their quantity and quality increasing. The current evolution of image generation techniques makes the detection of manipulated content through visual inspection increasingly difficult. This challenge has motivated researchers to analyze heart-beat-related signal to distinguish deep fakes from genuine videos.”

And we thought deep fakes couldn’t mimic those seemingly foolproof signals. We are wrong, according to a new international study published in the journal Frontiers in Imaging: “High-Quality Deepfakes Have a Heart!” authored by scientists from Fraunhofer Heinrich-Hertz-Institute HHI, and from Humboldt University, both in Berlin.

They analyzed deepfake videos of the faces of twelve participants with photoplethysmography, that looks at the blood flow changes reflecting heartbeat signal—the subtle color variations in our skin that are invisible to the naked eye. Such methods were considered to be extremely promising for identifying deepfakes. No longer!

The team recorded videos of twelve diverse participants under controlled conditions with uniform lighting. During each session, participants performed various activities like talking, reading, and interacting with the recording supervisor. Crucially, the researchers also measured the actual heart rates of selected participants using electrocardiogram and photoplethysmography sensors to normalize the data. Using these recordings, the researchers generated their own high-quality deepfakes using multiple methods, including employing an open-source tool called DeepFaceLive, which performs real-time face swapping. (For additional testing, the team incorporated deepfakes from public datasets, too.)

But therein lay the problem. Apparently high-quality deepfakes unintentionally retain the heartbeat patterns from their source videos, undermining traditional detection methods that relied on detecting subtle skin color changes linked to heartbeats. The fake videos, including their own creations, contained valid heart rate signals that closely matched those in the original source videos.

That means our ability to identify fake videos is failing. This challenges many existing detection tools, which may need significant redesigns to keep up with the evolving technology. To effectively identify high-quality deepfakes, researchers suggest shifting focus from just detecting heart rate signals to analyzing how blood flow is distributed across different facial regions—a unique and idiosyncratic distribution specific for a particular individual—thus providing a more accurate detection strategy.

Bottom line: Deepfakes are becoming more sophisticated, and it’s going to be a challenge to have detection systems be equally, if not more, sophisticated than the deepfakes themselves.

But the good thing is that God, who is not deceived, cares for, watches, saves, blesses, and guides our hearts.

You have put joy in my heart
more than when their grain and new wine abound.
In peace I will both lie down and sleep,
for You alone [are] Yahweh.
Psalm 4:7–8

But I—in Your lovingkindness I have trusted;
my heart will jubilate in Your deliverance.
Psalm 13:5

The innocent of hands and the pure of heart, …
He shall take away a blessing from Yahweh, ….
Psalm 24:4–5

With all my heart I have sought You; …
In my heart I have treasured Your utterance
in order that I may not sin against You.
Psalm 119:10–11

I have inclined my heart to do Your statutes,
forever, to the end.
Psalm 119:112

No deepfaking that!

SOURCE: Study Finds; Frontiers in Imaging