Email Indirect
A challenge to fool an LLM into leaking calendar information.
Learn
AIV Learn surfaces the public AI Village workshop material: hands-on labs for people who want to explore AI security by doing, not only by reading.
This is not a complete curriculum. It is a clear entry point into the open workshop infrastructure that already exists and can grow into more structured learning paths.
Workshop Philosophy
"get them excited to learn"
AIV's workshop philosophy is not that every participant masters the skill in the room. The goal is to make the topic accessible, spark interest, and give people a path to continue learning.
Source: AIV Workshops contributor guidelines.
Workshop Material
Prompt Extraction
A challenge to extract a secret hidden in an LLM's system prompt.
RAG Poisoning
A challenge to poison a RAG knowledge base and make an AI assistant spread misinformation.
Additional workshop directories in the public repository: LLM Embeddings, YOLO L2.
How Workshops Run
The public workshop repository describes containerized workshops. Each workshop is expected to serve a single website, with per-user pod environments and a Rust deployment/proxy approach. LLM-backed workshops depend on configured LLM service environment variables.
The repository also notes practical constraints: workshops should be self-contained, include a `docker-compose.yml`, and fit resource limits.
Experience Them
AI Village event pages include talks, demos, and workshops where that information is available in current site content. Start with the events archive and join the Discord for community discussion.
Contribute
Workshop contributions should be self-contained containers connected by `docker-compose.yml` and serving a single website. Review the public workshop repository before proposing new material.
What's Next
The next step is editorial: organize workshop material into clearer learning paths without overstating it as a formal training platform.