AI-Powered Social Engineering and Agent Security Threats

A new article demonstrates how generative AI enables sophisticated social engineering—deepfake job applicants, synthetic personas—with Amazon blocking 1,800 attempts. Combined with prior incidents of rogue agents and AI-to-AI attacks, this highlights the urgent need for AI-native security and trust infrastructure. The attacker speed advantage (63% CISOs agree) and 88% attacker vs 12% defender gap persist. A new practical guide on assessing sensitive data leakage risk reinforces the security theme.