← Frameworks
LLM07:2025

System Prompt Leakage

How LLM07:2025 System Prompt Leakage shows up in practice: the mapped risk classes in this atlas, the documented incidents that prove it's real, and the scenarios and controls to learn and defend against it.

Real-world cases

5

Documented incidents, disclosed vulnerabilities and research that illustrate LLM07:2025 — latest first, each with sources.

Browse all real-world cases →

Controls & guardrails that address this

5

Guardrails across the risks mapped to LLM07:2025, grouped by control function. Filter by control category below.

Control category
Preventive · 2
Instruction hierarchy / privileged system promptinteractive

Training the model to treat the app's standing instructions as more authoritative than anything a user or document says.

Open the Control Library →

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 ↗