The Most Interesting Thing in A.I. Deepfakes and Deception - with Nicholas Thompson and Eva Galperin
Dec 11, 2024
In this engaging discussion, Eva Galperin, the Director of Cybersecurity at the Electronic Frontier Foundation, delves into the dark side of AI. She tackles the alarming rise of voice-cloning scams and non-consensual imagery, stressing the urgent need for legal protections and societal change. Eva shares personal insights into her career in cybersecurity, revealing how technology can both empower and threaten our privacy. With a focus on mitigating risks and preserving truth, she highlights the critical role of encryption in fighting evolving digital threats.
AI Snips
Chapters
Transcript
Episode notes
Deepfake Dangers
- AI's role in deepfakes is often overstated; truly convincing fakes require close attention and post-production.
- The danger lies not in perfect replication, but in people's willingness to believe fakes aligned with their biases.
Spotting Deepfakes
- AI-generated images, even as they improve, will likely retain subtle inconsistencies or "tells."
- These tells, often uncanny elements, help identify fakes, although they evolve over time.
The Doctorow Scam
- Cory Doctorow, a self-proclaimed "internet guy", fell for a scam despite his expertise.
- The scammer posed as his bank, requesting his credit card details, exploiting a time of vulnerability and poor bank service.