Episode 5: Why AI agents break Zero Trust

June 9 | 10 a.m. PDT | 12 p.m. BST | 3 p.m. AEST

30 minutes

Traditional Zero Trust was built for predictable patterns of human users - patterns that AI agents have now effectively broken. While legacy security relies on anticipating user actions, agents determine their own requirements on the fly and vanish in seconds. As recent "goal-hijacking" incidents prove, the risk isn't just excessive permissions—it’s a failure of identity provenance.

Join Harish Peri (Okta) and Chris Hughes (Resilient Cyber) for a deep dive into how to secure the AI supply chain by prioritizing identity provenance over simple activity monitoring.

What you’ll walk away with:

Strategies to grant agents real-time, task-specific access instead of excessive permanent permissions.
Methods to map every autonomous identity to a human stakeholder for clear accountability.
The expertise to replace permanent credentials with short-lived tokens for transient agents.

Episode 5: Why AI agents break Zero Trust

June 9 | 10 a.m. PDT | 12 p.m. BST | 3 p.m. AEST

Join Harish Peri (Okta) and Chris Hughes (Resilient Cyber) for a deep dive into how to secure the AI supply chain by prioritizing identity provenance over simple activity monitoring.

What you’ll walk away with:

Strategies to grant agents real-time, task-specific access instead of excessive permanent permissions.
Methods to map every autonomous identity to a human stakeholder for clear accountability.
The expertise to replace permanent credentials with short-lived tokens for transient agents.

Agentic AI offers big potential but comes with real risks. Join us to talk about how to balance innovation with security, data and cost_ including what we’ve learned from Okta’s own AI journey.