EPISODE · May 19, 2026 · 51 MIN
Agents need boundaries | Fotis Chantzis from OpenAI
from Zero-Shot Learning · host Nancy Wang, Dev Tagare, Fotis Chantzis, OpenAI, 1Password, Google Gemini
In this episode: Old identity models weren’t built for agents The hardest security problems begin after access is granted Long-running agents expose the limits of OAuth Secrets remain one of the weakest links in agent design How to stop agent overreach in real time with short-lived credentials and runtime boundaries Zero-Shot Learning is a builder-to-builder podcast about how AI systems are designed, deployed, and secured. Subscribe for more. Go deeper: Episode companion blog How to protect against OAuth-based supply chain breaches and credential sprawl 1Password Developer newsletter Fotis: LinkedIn 1Password: 1Password.com 1Password Developer newsletter - sign up for updates to our developer tooling Build securely with 1Password Developer
Embed this episode
What this episode covers
Agents don't fit old identity models. As OpenAI’s Agent Security Lead, Fotis Chantzis has a front-row seat to see how agents push identity systems beyond what they were built to control. That’s where things start to fall apart and where most teams lose control. In this episode, 1Password CTO Nancy Wang and 1Password VP of Engineering for Developer and AI, Jeff Malnick, ask Fotis where authentication starts to strain under the actions of agent identities, why secrets handling can get risky fast, and how teams should think about dynamic access controls.
NOW PLAYING
Agents need boundaries | Fotis Chantzis from OpenAI
No transcript for this episode yet
Similar Episodes
Feb 4, 2026 ·18m