This repository accompanies the labs from the book **Secure AI Systems** and demonstrates both vulnerable and hardened AI architectures through practical security-focused implementations.
-
Updated
May 20, 2026 - Python
This repository accompanies the labs from the book **Secure AI Systems** and demonstrates both vulnerable and hardened AI architectures through practical security-focused implementations.
FUSE-based encrypted overlay filesystem for zero-trust cloud storage
Zero Trust lab in three layers, all as code: Microsoft Entra ID identity and Tailscale networking (ACLs, SSH, high-availability subnet routing) in Terraform, plus TypeScript AI agents (Scarlet coordinator, JML, GTM) with per-agent identity, human approval and full traces. Threat modeled against the OWASP LLM and Agentic Top 10. Mutation-tested CI.
To associate your repository with the zerotrustapplication topic, visit your repo's landing page and select "manage topics."