Inside the Strategy Room 288. Securing agentic AI: A playbook
49 snips
Jan 29, 2026 Charlie Lewis, cybersecurity and tech resilience lead with board-level advising experience, and Rich Isenberg, former CISO turned resilience partner, unpack agentic AI risks and governance. They define agentic versus generative AI. They cover attacker use of adaptive agents, operational fragility from shared models, risk-based agent classification, observability and kill switches, and treating agents like digital employees for safe scale.
AI Snips
Chapters
Transcript
Episode notes
Agent Risks Rise Without Massive New Budgets
- CISOs rank agent risks among their top three concerns but budgets are shifting, not exploding.
- Charlie Lewis notes spend moves toward identity and GRC rather than large net increases.
Use AI To Defend Against AI-Powered Attacks
- Deploy AI in defenses because attackers use agentic AI to adapt malware in real time.
- Rich Isenberg warns human-only SOCs cannot keep up with automated adversary adaptability.
Shared Components Create Fragility
- Shared models, centralized policies, and orchestration hubs create single points of failure.
- Rich Isenberg warns subtle data poisoning can cascade across procurement, finance, and operations.