Risk-based authentication checks sensitive actions as they happen, while narrow permissions limit what AI agents can do on a user's behalf.
The 2026 findings also highlight Docker and Kubernetes usage, interest in alternative infrastructure tools and limited AI adoption for production operations.