The Deepfake Supply Chain: Weaponizing Synthetic Media into Modern Cybercrime Operation
summary
Claim: synthetic-media fraud is a $40B problem with ~36% yearly growth and turnkey offerings (cloned videos $300–$800, full business-email packages $2,000) that defeat single-model detectors (accuracy falls to ~60% on unseen data). Implements a pipeline that extracts LFCC/constant‑Q/cepstral features for audio, uses FNet/ViT and other models for images, and scores an assembled verdict plus provenance checks with ED25519 cryptographic watermarking. Uses multimodel ensemble with attention disagreement maps and integrates alerts into SOC tooling, reporting 47% faster triage and <3s end‑to‑end response.
AI-written summary · about
chapters (10)
- 0:00Deepfake cybercrime threats
- 3:18Deepfake fraud ecosystem
- 8:20Deepfake supply chain attacks
- 13:41AI defense against synthetic media
- 20:06Voice clone attack detection
- 27:10Deepfake video liveliness bypass
- 30:33C2PA image authenticity
- 35:52Authenticity framework for SOC
- 38:39Evolving threats and defenses
- 40:46Closing remarks
related talks
30mReality Under Attack: Just How Serious Is the Deep Fake Threat?
30mAI-Generated Synthetic Media, aka Deepfakes and its impact
27mTruth or Fiction The Art of Unveiling DeepFakes in Text, Audio, Image
26mDeepfakes, audiofakes, and the future of media
27mIs now the time to solve the deepfake threat?
51mThe Great Brain Robbery: Navigating the Dark Future of Online Manipulation
41mDeepfake Threats in Business: What You Need to Know
28mMedia Provenance as a Prevention for Malicious Synthetic Media