Akamai: Half of Enterprise AI Bypasses Security

A new report from Akamai reveals that nearly half of enterprise AI use bypasses traditional security. This means a lot of AI activity goes unmonitored by security teams. The report highlights a rise in "shadow AI," where unmanaged applications operate silently. It also points to "AI power users" who drive most of the enterprise AI exposure. What's interesting is the emergence of new AI-native attack vectors. Researchers discovered three novel threat methodologies in 2026. These include "Vibe hacking," which manipulates coding assistants, "CursorJacking," where rogue browser extensions harvest sensitive data, and "CometJacking," which uses malicious instructions on web pages to compromise local AI agents. These new threats bypass traditional perimeter defenses entirely. The bottom line: rapid AI integration has outpaced security, exposing critical corporate assets to new cyber risks.