Anthropic's Claude Mythos has collapsed the time it takes to turn a vulnerability into a working attack, and the underlying capability is already spreading to cheaper, open-weight models. This is happening at the same moment AI agents are multiplying the attack surface. Agents authenticate, hold secrets, and act on their own, generating exactly the kind of privileged identity exposure an AI-driven attacker is built to find. Credentials are still the prize. What's changed is how fast they can be reached.
In this on-demand session, join to hear Delinea security leaders Jason Mitchell and Brian McHenry share five best practices they apply inside Delinea and recommend to identity and security leaders.
Attendees will learn:
- Why standing privilege is the attack vector and how just-in-time access removes it
- How to govern secrets, sessions, and runtime authorization before an attacker can escalate
- Why AI agents require the same identity discipline as human admins and how to apply it
- Where to start: an honest look at where standing privilege still lives in your environment
