- 0:00 - Intro: Black Hat 2026 and AI Agent Security
- 0:22 - Delinea’s Runtime Authorization for AI Agents
- 01:12 - Why AI Agents Need Action-by-Action Authorization
- 03:13 - Preventing Unauthorized Actions by Authorized Agents
- 04:11 - Tracking AI Agent Identity and Audit Trails
- 05:16 - Why AI Agents Behave Differently Than Humans
- 06:04 - The Growing Scale of AI Agent Identities
- 07:52 - Just-in-Time Credentials for AI Agents
- 09:03 - MCP Servers, Direct Connections, and Visibility Gaps
- 11:38 - How CISOs Should Approach AI Agent Security
- 12:50 - Securing AI With AI and AI Governance
- 14:30 - Machine Identities vs AI Agent Identities
- 15:13 - Closing Thoughts: The Future of AI Security
Frank Vukovits is Chief Security Scientist at Delinea and a 30-year experienced auditor, security, and business software professional. As a user, Frank has implemented and managed numerous ERP projects and now combines his software, security, and audit knowledge in his role at Delinea and is a frequent speaker at user groups and audit events. Frank has held corporate IT audit roles, and he carries Certified Internal Auditor and Certified Information Systems Auditor designations. Previously he was also a CISSP. He has presented educational sessions publicly for the past two decades and has been on all sides of auditing, as a user, an auditor, and a vendor of audit and security software in his role currently at Delinea.