{"type":"rich","version":"1.0","provider_name":"Transistor","provider_url":"https://transistor.fm","author_name":"Cyber Sentries: AI Insight to Cloud Security","title":"Securing AI Agents: How to Stop Credential Leaks and Protect Non‑Human Identities with Idan Gour","html":"<iframe width=\"100%\" height=\"180\" frameborder=\"no\" scrolling=\"no\" seamless src=\"https://share.transistor.fm/e/628d6eab\"></iframe>","width":"100%","height":180,"duration":1961,"description":"Bridging the AI Security Gap—Inside the Rise of Non‑Human IdentitiesIn this episode of Cyber Sentries from CyberProof, host John Richards sits down with Idan Gour, co-founder and president of Astrix Security, to unpack one of today’s fastest-emerging challenges: securing AI agents and non-human identities (NHIs) in the modern enterprise. As companies rush to adopt generative-AI tools and deploy Model Context Protocol (MCP) servers, they’re unlocking incredible automation—and a brand-new attack surface. Together, John and Idan explore how credential leakage, hard-coded secrets, and rapid “shadow-AI” experimentation are exposing organizations to unseen risks, and what leaders can do to stay ahead.From Non‑Human Chaos to Secure‑by‑Design AIIdan shares the origin story of Astrix Security—built to close the identity-security gap left behind by traditional IAM tools. He explains how enterprises can safely navigate their AI journey using the Discover → Secure → Deploy framework for managing non-human access. The conversation moves from early automation risk to today’s complex landscape of MCP deployments, secret-management pitfalls, and just-in-time credentialing. John and Idan also discuss Astrix’s open-source MCP wrapper, designed to prevent hard‑coded credentials from leaking during model integration—a practical step organizations can adopt immediately.Questions We Answer in This EpisodeHow can companies prevent AI‑agent credentials from leaking across cloud and development environments?What’s driving the explosion of non‑human identities—and how can security teams regain control?When should organizations begin securing AI agents in their adoption cycle?What frameworks or first principles best guide safe AI‑agent deployment?Key TakeawaysStart securing AI agents early—waiting until “maturity” means you’re already behind.Visibility is everything: you can’t protect what you don’t know exists.Automate secret management and avoid static credentials through just‑in‑time...","thumbnail_url":"https://img.transistorcdn.com/Ipg5CALrzv7pPIJnV_OJHWbm0TRU-H5nEd05XgrZwY8/rs:fill:0:0:1/w:400/h:400/q:60/mb:500000/aHR0cHM6Ly9pbWct/dXBsb2FkLXByb2R1/Y3Rpb24udHJhbnNp/c3Rvci5mbS9lZDg5/MzE1MjJkODgxYmJh/MzE2ZDA1ZjI5YmNj/YTM3OC5qcGc.webp","thumbnail_width":300,"thumbnail_height":300}