PromptLab
Sandboxed

Learn AI attacks safely.
Test defenses live.

PromptLab is an interactive AI security lab for exploring prompt injection, jailbreaks, and LLM defense strategies through real-time sandbox simulations.

Sandboxed

No external systems contacted

227 Techniques

Across 18 attack families

Learn by doing

Attack, then defend, then understand

Available Scenarios

Each scenario is an isolated sandbox target with a specific vulnerability to explore.

Loading scenarios...