← Back to work

ai security

AI Security Course

PhD-grade red-team curriculum — open source.

Overview

Twelve hands-on modules spanning adversarial CNN attacks, audio attacks, LLM prompt injection, jailbreaks, model extraction, embedding inversion, and data poisoning — runnable locally with no API costs.

Highlights

  • 12 modules
  • Local-first
  • DARPA-rooted

Stack

Adversarial MLLLM SecurityRed TeamingDocker