SECURITYWEEK reports that social engineering remains a highly effective attack vector, with trends pointing to increasing sophistication thanks to AI-generated deepfakes and related tricks. Traditional user training is criticised as lacking robust evidence of real-world effectiveness, leaving organisations seeking alternatives. The piece highlights a shift towards technology-assisted detection of live social engineering, alongside or in place of human vigilance, as attackers target human psychology in real time.
Netarx is presented as an early example of this tech-forward approach. Its AI defence meta harness continuously scans communications using more than 40 AI models and over 1,000 signals, including device fingerprints, EXIF data, compression signatures, and location checks, to detect whether a feed is authentic or potentially synthetic. For AI-generated voice or video, the system looks for artefacts such as unnatural silences, microlayer irregularities, lip-sync mismatches, and micro-expression anomalies.
A colour-coded traffic light—green for verified, amber for unknown or suspicious, red for confirmed deepfake—guides user action in real time, designed to inform rather than automatically block. The solution runs on Windows, macOS, Chrome, iOS and Android, and is accompanied by an Impact Database of real-world incidents. The article notes notable incidents in Las Vegas venues and a Brinks Home breach to illustrate evolving risk.