Deepfakes and the Crisis of Evidence

Generative AI can now fabricate convincing images, voices, and video. Two dangers follow. The obvious one is forgery — fake evidence used to deceive. The subtler, larger one is the 'liar's dividend': once everyone knows media can be faked, the powerful can dismiss *genuine* evidence as fake. The shared baseline of 'pics or it didn't happen' erodes, which is exactly the condition Arendt warned about — a public that can no longer tell true from false becomes easy to rule. Practical responses are emerging but imperfect: provenance and content-authentication standards, cryptographic signing at capture, and old-fashioned source verification (who published this, with what track record, corroborated by whom?). This entry is labeled speculative where it forecasts: the social equilibrium is genuinely unsettled, and humility about the trajectory is part of honest analysis.