
Researchers at the University of California, Los Angeles, have developed a new technique to detect deepfake videos instantly by analyzing the light distributed within them.
Recent studies indicate that humans are only able to spot deepfakes about 50% of the time despite obvious visual anomalies like strange eye movements or extra fingers.
In laboratory tests, the new method proved exceptionally reliable, succeeding almost every time in identifying manipulated video content.