Uploaded August 2026 | Updated November 2026, 3 hours ago
π€ Check out Okta here: https://ul.live/okta_yt
π I had a great chat with Harish about the wild west of AI agents and why securing their access is basically just classic identity security applied to a brand-new surface. We dug into how things like Okta's agent gateway act as a much-needed bouncer for these bots, letting devs move fast without accidentally handing an AI the keys to the entire kingdom.
π§ What we talk about:
π€ The AI Agent Explosion: The rapid deployment of AI agents across enterprises and why their unchecked access to sensitive production data is keeping security teams up at night.
π§± Back to Basics: Why the solution to securing AI agents isn't actually new, but rather a return to classic identity, authorization, and security fundamentals.
π§Ύ The Auditing Nightmare: The impending compliance challenges of proving the exact chain of custody from a user's initial prompt all the way to an autonomous agent's specific transaction.
π Centralized Control and Kill Switches: How tools like Okta's agent gateway act as an independent control plane to manage resource connections, issue tokens, and execute an organizational kill switch if things go wrong.
βοΈ Friction vs. Automation: The emerging threat of agent-to-agent scope escalation and why heavily relying on "human-in-the-loop" safeguards completely defeats the speed and purpose of AI
π Stay connected
π¬ Subscribe to the newsletter at:
danielmiessler.com/subscribe
π¬ Join the UL community at:
danielmiessler.com/upgrade
π£ Follow on X:
https://x.com/danielmiessler
πΌ Follow on LinkedIn:
linkedin.com/in/danielmiessler
π€ Check out Okta here: https://ul.live/okta_yt
π I had a great chat with Harish about the wild west of AI agents and why securing their access is basically just classic identity security applied to a brand-new surface. We dug into how things like Okta's agent gateway act as a much-needed bouncer for these bots, letting devs move fast without accidentally handing an AI the keys to the entire kingdom.
π§ What we talk about:
π€ The AI Agent Explosion: The rapid deployment of AI agents across enterprises and why their unchecked access to sensitive production data is keeping security teams up at night.
π§± Back to Basics: Why the solution to securing AI agents isn't actually new, but rather a return to classic identity, authorization, and security fundamentals.
π§Ύ The Auditing Nightmare: The impending compliance challenges of proving the exact chain of custody from a user's initial prompt all the way to an autonomous agent's specific transaction.
π Centralized Control and Kill Switches: How tools like Okta's agent gateway act as an independent control plane to manage resource connections, issue tokens, and execute an organizational kill switch if things go wrong.
βοΈ Friction vs. Automation: The emerging threat of agent-to-agent scope escalation and why heavily relying on "human-in-the-loop" safeguards completely defeats the speed and purpose of AI
π Stay connected
π¬ Subscribe to the newsletter at:
danielmiessler.com/subscribe
π¬ Join the UL community at:
danielmiessler.com/upgrade
π£ Follow on X:
https://x.com/danielmiessler
πΌ Follow on LinkedIn:
linkedin.com/in/danielmiessler










