Deepfake Audio and Video in Social Engineering Attacks: Threat Assessment and Detection

What Did This Cybersecurity Research Find?

This cybersecurity threat assessment studied how deepfake audio and video are being used in social engineering attacks targeting organizations. Cybersecurity defenses against deepfake-enhanced social engineering remain weak, with participants in controlled experiments detecting deepfake audio at rates barely above chance (54%) without specialized training.

Key Findings

• 1Untrained participants detected deepfake audio at only 54% accuracy (barely above chance)
• 2Deepfake-enhanced vishing (voice phishing) succeeded at rates 3.2x higher than traditional vishing
• 3Training on deepfake cues improved detection accuracy to 78%
• 4Real-time deepfake video in video calls was detectable through latency and artifact analysis
• 5Organizations with verbal verification protocols (challenge-response codes) blocked 92% of deepfake voice attacks

How Does This Apply to Cybersecurity Careers?

Security awareness professionals need to prepare for deepfake social engineering. Incident responders should understand this emerging threat vector and its detection challenges.

Who Should Read This?

Frequently Asked Questions