What practice secures AI agents in enterprise environments?

Michael Jenkins

C.T.O. · Threatlocker

Zero trust

Michael Jenkins says his team uses a Zero Trust policy and ring fencing to limit what users and AI can access, because AI can do a lot but the company must decide what it should be allowed to do.

Jason Elrod

Chief Information Security Officer (CISO) · Multicare Health System

Understand identities and permissions

Jason Elrod says leaders should understand the identities, permissions and controls they already have, calling that key, and start by separating agents operating under someone else's identity before buying another AI security product.

The answer starts at 15:03.

Mandy Andress

CISO · Elastic

Inventory AI usage

Mandy Andress says the first priority is to get an inventory of AI usage, including agentic AI, because if you don't know where it's being used it's much harder to put controls and structure around it.

The answer starts at 15:01.

Jacob Thomas

Manager of Information Security · Texas Children'S Hospital

Good governance model

Jacob Thomas says his approach is to have a good governance model behind everything, using mitigating and compensating controls to reduce the attack surface — though he says this isn't unique to AI.

JA

James J Azar

Chief Information Security Officer · Artera Services

Continuously validate AI behavior

James J Azar says continuously validating AI behavior should become a standard architectural requirement, not an optional safeguard, citing autonomous AI systems that can identify exposed infrastructure, chain vulnerabilities, and move laterally faster than most organizations can detect.