Deepfake drills expose verification gaps in 62% of firms hit by voice clones

The evidence trail starts with public audio from earnings calls and webinars that trains voice models in hours. Osterman Research January 2026 findings confirm 60% of security leaders lack confidence in current training, shifting reliance to repeatable verification steps rather than auditory detection. FBI May 2025 guidance already directs recipients to call independently sourced numbers, yet incident reports show help desks and identity teams remain the weakest nodes for MFA resets and admin account creation.

Process failures surface when urgency from a cloned CFO overrides callback protocols. The drill maps executives whose voices are most exposed alongside staff who control wires or access, revealing that standard policies survive tabletop exercises but collapse under real-time pressure from a familiar voice.

Next steps require quarterly live drills that insert synthetic calls into actual workflows, measuring whether independent verification occurs before any action. Procurement records show rising demand for out-of-band confirmation tools tied to identity platforms, indicating the pattern will accelerate as real-time face swaps improve.