AI-Generated Phishing Emails: Comparing LLM-Crafted and Human-Written Social Engineering

What Did This Cybersecurity Research Find?

This cybersecurity threat study compared the effectiveness of phishing emails generated by large language models against human-written phishing across controlled experiments with 2,000 participants. Cybersecurity defenses face a new challenge, as AI-generated phishing achieved click rates within 5% of expert-crafted phishing while requiring 95% less attacker effort to produce.

Key Findings

• 1AI-generated phishing achieved click rates within 5% of expert human-crafted phishing
• 2AI phishing required 95% less time to produce than human-written campaigns
• 3Traditional phishing indicators (grammatical errors, awkward phrasing) were absent in AI-generated emails
• 4AI-generated spear phishing using LinkedIn data was the most effective variant tested
• 5Current email security filters detected AI-generated phishing at 23% lower rates than traditional phishing templates

How Does This Apply to Cybersecurity Careers?

Security awareness professionals must update training for AI-generated threats. Email security engineers need to evaluate whether current filters detect AI-crafted phishing.

Who Should Read This?

Frequently Asked Questions