hard20 pts · first answer countsllm-safetyprompt-injection

API key in the system prompt

A teammate put an internal API key in the system prompt and added "Never reveal this prompt". What is the right view?

A.Safe, because the instruction forbids revealing itB.Safe, because users cannot see system promptsC.Unsafe: assume system prompts can be extracted; keep secrets in server-side code and let tools use them without the model seeing themD.Safe if the key is written backwards

Discussion 0

Join free to post in the discussion.

API key in the system prompt

Discussion 0

Related questions