March 2026 brought numbers to what fraud teams have felt for two years. A survey of more than 700 anti-fraud professionals across eight regions found deepfake social engineering rising at the sharpest rate of any fraud type, with 77% of respondents reporting increases, and generative document forgery and deepfake injection close behind. The most sobering figure: only a single-digit percentage of organisations describe themselves as firmly prepared.
Preparedness is not vigilance
The unprepared majority is not lazy; it is armed with the wrong tools. Training staff to "spot" deepfakes is a losing race against models that improve monthly. Detection AI helps but plays defence on the attacker's field: perception. Preparedness that survives contact means moving decisions off perception entirely.
The prepared minority shares an architecture
Organisations reporting resilience cluster around the same design: government-signed credentials instead of document photos, certified liveness (ISO 30107-3) that catches injection at the sensor level, identity-gated video for high-stakes conversations, and cryptographic sealing of records so tampering is evident. That stack is WeVerify's daily business: chip-level KYC, liveness and deepfake detection, sealed video calls. The survey's gap between the 77% threat and the single-digit readiness is precisely the market's to close this year.