Recorded live and available on demand: Securing Today’s AI Agents: What Your Team Needs to Know | Cybersecurity Dive

Speakers/Host: Andrew Alaniz (CipherNorth) joined by Jeremy Kirk from Okta.
Hosted & Sponsored by:Okta and Banking Dive

I joined Jeremy Kirk, with Okta, to talk about securing AI agents. The biggest take away is that while Generative AI technology is rapidly evolving, cybersecurity requirements are staying the same. AI just reinforces the need to get the basics right.

• Use unique identities - we’ve been assigning IDs to servers, scripts, and systems for years. AI Agents are no different.

• Log relevant actions - Just like we need to know who does what, when. We need to know the same things about AI Agents.

• Use least privilege - We had the issue for years where access was hard, so it was easy to assign admin permissions. Don’t fall into that trap with AI Agents. They are too capable and can cause a lot of damage.

• Know where your data is and who has access to it - If you want to limit the damage an agent can cause, limit what it can access.

Watch the full recording to hear us discuss some areas where business should focus as they begin to deploy AI Agents. Deploying AI agents and need help getting the governance right? Learn about our GenAI Security Planning services.