Aug 17, 2026 · 46 min
Deepfakes expose the limits of online trust
The Future of Deepfakes and the Decline of Reality with Hany Farid
Synthetic media now enables fraud, reputational attacks, and non-consensual sexual imagery faster than individuals can reliably detect or contest it.
- 1PhotoDNA showed how image authentication could identify known abusive material at scale without reconstructing originals.
- 2Deepfakes have become practical tools for scams, reputational attacks, and non-consensual sexual imagery, with platforms enabling their spread.
- 3Human perception is unreliable against sophisticated manipulation, making automated detection, investigative context, and stronger platform accountability essential.
Don't miss
Farid explains why even experts struggle to identify sophisticated manipulation visually, and why computational tools and investigative context matter.
The brief
Hany Farid traces digital-image authentication from PhotoDNA, developed to identify known abusive images at scale, to modern systems for detecting manipulated and AI-generated media.
The threat has moved from theory to practice: deepfakes now support scams, attacks on reputations, and non-consensual sexual imagery, while ordinary targets lack comparable resources.
Farid argues that platforms and major technology companies should bear more responsibility, especially as nudify apps reach schools and target young teenagers.
Visual judgment is not enough: even experts can miss subtle manipulation, while computational tools combined with human expertise can expose what perception cannot.
The conversation ends with a provocative possibility: as digital evidence becomes less trustworthy, younger generations may revive analog forms of communication and proof.
Featuring
Hany FaridMentioned
Mark Elliot Zuckerberg
Elon Reeve MuskListen to the full episode and explore every guest, topic, and moment on PodLume.