RECORDING
Talk recap: threat-informed guardrails for AI-assisted code (DEF CON AppSec Village)
Recording of our DEF CON AppSec Village talk on why taxonomy-only guardrails fall apart for AI-generated code, and what to replace them with.
WhenAugust 9, 2026 at 8:00 PM PDT
WhereDEF CON 34, Las Vegas
Speaker Abhay Bhargav, Founder, VibeReview
Talk recap: threat-informed guardrails for AI-assisted code (DEF CON AppSec Village)
Why a SAST rule that fires on Django will not fire on the same vulnerability when an agent rewrote the route last week. The talk walks through three live demos: a taxonomy-only guardrail set on an evolving FastAPI repo, the same set with threat-informed context, and the audit-log diff between the two.
Resources from the talk
- Slides on the VibeReview GitHub
- Demo repos with starter threat models
- Companion explainer in /learn