"people can accurately spot whether an image, audio or video is genuine about 65 per cent of the time (chance is 50 per cent)" [65 per cent]
"technology has democratised deception: it takes only motivation and a mobile phone to churn out convincing deepfakes" [mobile phone]
Farid's lab data shows humans cannot reliably distinguish real from synthetic content. The article then traces how this perceptual collapse cascades: verification takes days (Iran war example), institutions must adopt costly forensic tools, and the burden shifts to consumers. This creates a structural advantage for actors with resources to generate or amplify deception, while truth-seekers face escalating friction. The dynamic generalizes beyond any single deepfake incident to reshape information power itself.
The article frames deepfake generation as now accessible to any motivated actor with basic hardware. Gartner's forecast that one in four job applicants will be fake by 2028 illustrates the scaling effect: as production costs approach zero, attack surface expands exponentially. Combined with algorithmic amplification and chatbot intermediation, this creates a structural shift where deception production outpaces detection and verification, favoring volume-based campaigns over quality-based truth.