Generative AI Has Made Phishing Attacks Nearly Undetectable, Experts Warn

Generative AI and large language models have eliminated the telltale signs of phishing emails — such as poor grammar, generic greetings, and awkward phrasing — that cybersecurity training long relied upon to identify threats. Attackers can now use AI to produce flawless, localized content in any language, automate the scraping of personal and corporate data, and run behavioral tests to optimize deceptive messages before mass deployment. Voice-cloning technology adds another dimension, allowing criminals to replicate a person's voice from as little as 30 seconds of audio to impersonate executives in real-time calls. Security experts argue that urging employees to stay vigilant is no longer a viable defense, as even careful professionals can be deceived by highly convincing, contextually accurate lures. Organizations are being advised to shift toward process-driven, zero-trust verification systems rather than relying on human judgment to detect fraudulent communications.