86% of Phishing Attacks Are AI Driven, KnowBe4 Research Finds

Key Takeaways

• 86% of phishing attacks now use AI-generated content
• Calendar invite phishing up 49% in six months
• Reverse proxy usage to steal M365 credentials rose 139%
• Microsoft Teams attacks increased 41% year‑to‑date
• Multi‑channel attacks now target humans and AI agents

Pulse Analysis

Artificial intelligence has become the engine behind the majority of modern phishing campaigns, with KnowBe4 reporting that 86% of attacks now incorporate AI‑generated content. This capability allows cybercriminals to craft highly personalized messages at scale, reducing the time and effort required to bypass traditional email filters. As AI tools become more accessible, attackers can quickly adapt their lures to current events, making detection increasingly difficult for conventional security solutions.

Beyond the inbox, the report highlights a surge in alternative attack surfaces. Calendar invitation phishing grew 49% as remote work drives reliance on scheduling tools, while Microsoft Teams assaults rose 41%, exploiting the platform’s real‑time communication features. Reverse proxies, used to harvest Microsoft 365 credentials, jumped 139%, reflecting a sophisticated approach to bypass multi‑factor authentication. The shift toward multi‑channel orchestration—combining email, calendar, and collaboration apps—creates a layered threat that targets both human users and the AI agents they interact with.

For enterprises, the implications are clear: security programs must evolve beyond email‑centric defenses. Integrated threat intelligence, AI‑aware detection, and continuous security awareness training are essential to counteract these blended attacks. KnowBe4’s recent launch of an AI‑driven assessment agent illustrates a proactive response, enabling organizations to simulate realistic, multi‑vector phishing scenarios and reinforce user vigilance. Investing in comprehensive, adaptive security controls will be critical to mitigate the rising tide of AI‑enabled social engineering.