jailbreak arena · 10 levels

Try to make mesay what I shouldn't.

Each level has a target phrase I've been told not to say. Every prompt is piped through a real defender LLM with an escalating system prompt. Beat one to unlock the next. Public attempt count.

Heads up: I can read every winning prompt (yours included). I review them periodically, patch leaks into the defender, and add new levels. Harder over time isn't me being mean — it's the game evolving with the players.