An attack where malicious instructions are embedded within external data (such as web pages or emails) that the LLM processes, leading to unauthorized action execution.
Want to actually apply concepts like this instead of just reading definitions?
Practice free on Zamlom