Super Data Science: ML & AI Podcast with Jon Krohn 989: Security for Mythos-Era Agentic Risks, with Rubrik’s Anneka Gupta and Cal Al-Dhubaib
11 snips
May 5, 2026 
Cal Al-Dhubaib, Principal Technologist who builds trust engineering for AI, and Anneka Gupta, Rubrik’s Chief Product Officer focused on cyber resilience and AI governance, discuss AI-driven security risks. They cover how agentic systems find and exploit vulnerabilities, why zero trust and least-privilege matter, Rubrik’s Agent Cloud, SAGE policy enforcement, and Agent Rewind for remediation.
AI Snips
Chapters
Books
Transcript
Episode notes
Backup Evolved Into Cyber Resilience
- Rubrik reframed backup as cyber resilience after attacks shifted from accidents to deliberate breaches.
- Anneka Gupta explained attackers now cause data loss more often than natural disasters, so rapid restore and identity recovery became central.
Agentic AI Forces Assume Breach Posture
- Agentic AI speeds both discovery and exploitation of vulnerabilities, shifting security from detection to assume-breach recovery.
- Anneka Gupta emphasized Mythos-class models can find and exploit vulnerabilities at machine speed, requiring machine-speed response.
Test And Assign Who Recovers When AI Fails
- Prepare an explicit recovery plan before deploying AI: define who recovers, how, and test it.
- Cal Al-Dhubaib urged organizations to practice failover because observability and testing alone aren’t enough when models fail.
