eThinker: Beware the AI-Generated Deepfake

A few years ago, when the media was reeling from reports of a now discredited dossier that included the “Pee-Pee Tape” from Moscow, the technical world was also becoming alarmed about deepfakes, manufactured pictures and videos that were so realistic they could fool anyone. We predicted that, even if the Pee-Pee Tape didn’t exist, it would within a year.

We weren’t wrong. A late-night comedian aired a grainy–and totally fictitious–flick of the former president in the hotel room with the prostitutes. Cue the laughter–except this is not funny, and it has only gotten worse.

The visual deepfakes that have been produced since then, mostly for the purpose of embarrassing the target, are uncannily close to reality, and now comes the AI-generated voices that can fool all but the experts into believing, “Yeah, he really said that–the creep!”

Criminals are already using deepfake voice technology to extort ransom from frightened parents who hear and believe the cries of their supposedly kidnapped children. But its most dangerous use will be in the realm of politics, to sway the voters into believing things that are not, in the rational light of day, possible.

Voltaire wrote: “Those who can make you believe absurdities, can make you commit atrocities.” Those prophetic words came true less than a century ago when a little corporal named Hitler took over then-democratic Germany. And that was when the only tools in his arsenal were the spoken and printed word.