Akamai: 47% Enterprise AI Bypasses Security, New Risks Emerge

New cybersecurity risks are emerging as nearly half of all enterprise AI use bypasses corporate security. An Akamai report reveals that 47.11% of enterprise AI conversations happen through personal identities, not corporate accounts. This "shadow AI" is not approved, monitored, or audited by IT teams. What's interesting is that while AI use is growing, much of the risk comes from a small group of highly active "AI power users." The report also highlights new AI-native attack methods like vibe hacking, CursorJacking, and CometJacking, which can bypass traditional defenses. These methods involve manipulating AI coding assistants, collecting sensitive data via rogue browser extensions, or manipulating local AI agents through malicious instructions on public webpages. The bottom line is that traditional security tools built for an earlier era are struggling to keep up with how sensitive corporate information is now used with AI. This means organizations need to rethink their security strategies for AI.