Akamai report finds nearly half of enterprise AI use bypasses corporate security controls
AKAM•AI-native attack methods identified
It identified three 2026 AI-native attack methods that can bypass perimeter defenses, including extension-based credential theft, indirect prompt injection, and workflow manipulation.
Mitigation priorities for security teams
Mitigation priorities include forcing SSO, shifting from static DLP to prompt-level inspection, tightening extension governance, and applying least-privilege controls to autonomous agents.
Akamai warns on unmanaged enterprise AI use
Akamai published its Enterprise AI Usage Risk Report 2026, warning nearly half of enterprise AI use bypasses corporate security controls.
Risk is concentrated among “AI power users,” with about 5% of employees driving most interactive AI prompts.
The report flags a large “shadow AI” long tail of unmanaged AI SaaS tools, creating major visibility gaps for security teams.
