Sandbox

Build an AI system, watch the risks light up

Drop building blocks onto the canvas and connect them. Connections that don't make architectural sense are blocked with a reason. As your system takes shape, the relevant risks activate — split into what your components imply and what your wiring creates — each with the guardrails that address it. Use the depth dial (top-right) to tier every explanation from plain-language to expert.

Your systemUntrustedaskscontextreplyshows🧑User💬Chat / AppInterface🛡️Input Guardrail🧩Prompt Assembly🧠LLM🧯OutputGuardrail
InstructionsDataActionsControl / decisionFeedback / logs

Example: a minimal chatbot. This interactive playground lets you build any AI system from 9+ building blocks and see its risks activate. Risks implied by these components:

AI RiskAtlas is an educational model of how GenAI & agentic systems work and fail. Architectures and payloads are illustrative and simplified for learning — not operational guidance. Real-world cases are summarised from public reporting.

Sources & further reading →·Built by Shi Yuan ↗