Your newest insider is an AI agent

What we covered

• The Hugging Face incident: what happens when your own agent attacks someone else
• Why AI attackers leave no fingerprint, and who is accountable when an agent hacks
• Coding agents as the new insiders: your access, not your judgment
• Agents holding production credentials, and developers who auto-approve everything
• Generative AI in 2026: still the Wild West?
• Guardrails with business context, built into every coding agent your developers choose
• Why audit needs proof: an immutable record of what agents did
• What comes next: agent identity, liability and the human in the loop

Chapters

1. 0:00Welcome
2. 0:52Jason Lee: from Microsoft to Zoom and Splunk
3. 1:34Meet the CybeDefend founders
4. 3:04Hugging Face: when your own agent attacks others
5. 4:34Why AI attackers leave no fingerprint
6. 6:40Who is accountable when an agent hacks?
7. 7:33Coding agents are the new insiders
8. 8:03Securing AI at Zoom in 2020
9. 10:03Humans are leaving the loop
10. 12:16Agents holding production credentials
11. 13:10Generative AI: still the Wild West?
12. 14:25Guardrails with business context inside the agent
13. 16:28Supporting every agent your developers choose
14. 18:19Proof for audit: an immutable record
15. 19:40What comes next: agent identity and liability
16. 21:49Will humans stay in the loop?
17. 23:30Jason's question: where agent security goes next
18. 26:06Closing thoughts