Full Segment Notes
Identity Security now spans two extremes. On one end, decades-old infrastructure that was never designed to be secure. On the other, rapidly emerging AI agents and non-human identities that organizations are racing to deploy. Most security teams sit in the middle, managing expanding identity risk with a fragmented stack of point solutions.As AI adoption accelerates, identity risk is scaling just as quickly. And agentic security is, at its core, an identity problem. The more access an AI agent has to corporate systems, the more powerful—and more dangerous—it becomes. Without deep identity context, there is no way to make an informed, real-time decision about whether an agent’s action is legitimate or overreach.In this session, Ron Rasin examines how the industry’s missteps with Active Directory, service accounts, and machine identities created blind spots that still persist—and how those lessons must shape the way we secure agentic AI from the outset. Ron will outline why identity must serve as the control plane for AI-driven enterprises, and why enforcement must happen at runtime—before access is granted, not after damage is done. He will also share recent Silverfort innovations and partnerships with leading AI platforms to deliver embedded identity controls across human, non-human, and AI identities, including Copilot Studio agents, third-party agents, and hybrid environments.Because when AI begins authenticating at machine speed, reacting is no longer enough.Segment Resources:https://www.silverfort.com/press-news/silverfort-unveils-ai-agent-security/
https://www.silverfort.com/blog/4-ways-ai-agents-change-the-way-we-approach-identity-security/To learn more about Silverfort and our AI Agent product, visit us at https://securityweekly.com/silverfortrsac.Read the interview summary from SC Media here: Silverfort’s Ron Rasin on the identity crisis of agentic AI
Key Moments
- 0:00 - Welcome to RSAC 2026 with Silverfort
- 0:23 - What Silverfort Does in Identity Security
- 01:16 - Human vs Non-Human vs AI Identities
- 01:37 - The Identity Sprawl Problem
- 02:56 - Why Runtime Access Control Matters
- 03:16 - Securing Service Accounts with Virtual Fencing
- 03:43 - Least Privilege for Non-Human Identities
- 04:15 - The Risk of AI Using Human Credentials
- 04:32 - Mapping the Full AI Identity System
- 06:11 - Why Admin-Time Security Is Too Slow for AI
- 06:47 - Managing AI Risk Across Third-Party SaaS
- 07:37 - Silverfort’s Integrations with AI Platforms
- 07:47 - How the Copilot Studio Integration Works
- 09:06 - Policy-Based Runtime Control for AI Agents
- 09:54 - Catching Overprivileged AI at Runtime
- 10:36 - Reporting AI Permissions to Leadership
- 11:09 - Educating Developers to Build Secure Agents
- 11:31 - Treating AI Agents Like Human Identities
- 11:59 - Avoiding Old IAM Mistakes with AI Accounts
- 13:14 - Delegating Only the Right Privileges to AI
- 13:41 - Reducing AI Access to What It Actually Needs
- 13:56 - Solving AI Identity Risk at Runtime
- 14:22 - What Makes Silverfort Different
- 15:25 - Closing Thoughts and RSAC 2026 Coverage
Guest
As Chief Strategy Officer, Ron leads Silverfort’s strategic alliances with technology partners, as well as our growth operations and business strategy. He brings more than 15+ years of hands-on product management experience and cyber security expertise. Prior to joining Silverfort, Ron was the Director of Product Management at Claroty, and held product management roles at Wix and NCR. Before that Ron served as a Team Leader at the 8200 elite cyber unit of the Israel Defense Forces. Ron holds a B.A in Economics from Tel Aviv University.
