As deepfake technology grows more accessible, we need to get savvier at detecting AI-doctored videos and audios. Watch for these telltale signs of manipulated media
Due to technical limitations, deepfake movements often appear jittery and disjointed rather than smooth and natural. Subtle discrepancies in gestures and gait can expose manipulated footage.
AI often struggles to generate natural blinking patterns. Deepfaked eyes may rarely blink, blink out of sync, or blink unnaturally slowly/rapidly. Normal blinking cadence is hard to mimic.
Deepfake tools sometimes decouple facial expressions from speech, resulting in a mismatch between emotions displayed and words said. Watch if they align convincingly.
Due to the technical difficulty of accurately mapping mouth movements, deepfaked mouths tend to appear blurred or warped when animated. Teeth also look unnatural.
Deepfake skin tones often seem off, with inconsistent texturing and coloring. Skin may look too smooth, wrinkled, or just generally artificial upon close inspection.
With limited data on individuals, deepfakes cannot completely mimic normal mannerisms or speech patterns. Subtle but “off” discrepancies in conduct could be telltale signs.
Lighting and shadows probably won't sync perfectly between the deepfaked face region and the rest of the body and background environment.
Deepfakes are unable to recreate small details like moles and may contain visible digital artifacts from the AI process. Real people have random imperfections that AI misses.
Voices may sound slightly choppy, garbled or robotic without the proper cadence, inflections and fluidity of normal speech.
Source file metadata can reveal manipulations and tampering not evident visually. Scrutinize details like timestamps and editing software used.
Stay vigilant for any of these warning signs when assessing media authenticity. While deepfakes grow more sophisticated, telltale hints of manipulation remain. Learn to spot the “tells” of AI trickery