SN 1090: Black Hat - The Hidden Flaws in AI Security Nobody Saw Coming
Summary
In this Black Hat live episode, Steve Gibson, Leo Laporte, and Richard Lowe discuss their evolving relationships with AI, contrasting Steve’s cautious approach with Leo’s heavy use of agentic coding tools. They cover a "Zoox Storm" hack video, the shift toward local AI models for privacy, and AI’s impact on cryptography and software development. The hosts also address listener feedback on AI-assisted app creation and the challenges of defining bug-free code.
Topics discussed
Intro: Security Now Live at Black Hat with Guests
AI Coding, Agents, and Local vs. Cloud Models
Listener Stories: AI for Legacy Apps and Finance
AI, Cryptography, and the Mythos/Fable Controversy
AI Vulnerability Discovery vs. Patching Capabilities
Dual-Use AI, Safety Filters, and Distillation
AI in Education, Society, and the 'Spaghetti' Era
LLM Security Flaws, Neil Stevenson, and Outro
Listen ad-free on Castria