AI agent identity: how to govern agentic AI in 6 stages

At RSAC 2026, security leaders warned that AI agents can pass identity checks yet still make harmful changes inside enterprises.

Executives cited cases at major companies where agents altered sensitive settings, showing that valid logins alone are no longer enough.

Cisco introduced a six-stage maturity model for governing agent identities, shifting security from access verification to action-level enforcement.

The broader concern is that agentic AI can operate at machine speed, making damaging decisions without human judgment or tighter authorization controls.