Skip to content
#

ai-security-research

Here are 18 public repositories matching this topic...

Curated, vendor-neutral open-source tools, standards, benchmarks and labs for AI Security Engineering, organized by lifecycle (Discover, Design, Build, Break, Defend). From the AI Security Engineers Community.

  • Updated Oct 27, 2026
  • Python

Secure Harness AI — the security control plane for production AI agents. SHAI sits between your agent and everything it can touch: tools, external APIs, files, and users. It enforces security at the boundaries — not by hoping the model behaves, but by intercepting every action before it executes.

  • Updated Oct 30, 2026
  • Python

One poisoned document hijacks Google's Gemini 3.1 Pro agent: it scans your Drive, harvests secret-shaped keys, plants persistent artifacts, and sends everything out through Google's own fetcher — from Google IP space.

  • Updated Oct 24, 2026

What one node failure costs a request on exo, vLLM and llama.cpp RPC: detection, recovery and C measured in 107 trials, fed back into the distributed-inference resilience model. Experiment 3 of a series on detecting, containing and removing a model split across many machines.

  • Updated Oct 27, 2026
  • Python

Secure AI data and RAG pipeline demonstrating provenance validation, retrieval authorization, prompt-injection defenses, sensitive-data protection, & MITRE ATLAS threat mapping. Project 5

  • Updated Nov 8, 2026
  • Python

AI supply chain security lab covering model, package, container, and dataset inventories; dependency pinning; vulnerability scanning and triage; provenance records; artifact hash verification; rejection of altered artifacts; and release rollback, with pipeline diagrams and release checklists. Project 8

  • Updated Nov 8, 2026
  • Python

What a distributed-inference swarm looks like on the wire: tcpdump captures of llama.cpp RPC, exo and LocalAI P2P on a Podman bridge, readable byte, port and rhythm rules, 15 predictions scored against the measurements. Experiment 4 of a series on detecting and stopping a swarm nobody operates.

  • Updated Nov 3, 2026
  • Python

Add this topic to your repo

To associate your repository with the ai-security-research topic, visit your repo's landing page and select "manage topics."

Learn more