The security practice of submitting crafted queries to an LLM to discover system limits, prompt leakages, and bypass routes.
Want to actually apply concepts like this instead of just reading definitions?