Prompt injection: A visual, non technical primer for ChatGPT users
A dummy's guide the LLM's fatal flaw: Prompt Injection which kills the idea of agent decision making pretty thoroughly.
Prompt injection: A visual, non technical primer for ChatGPT users
What is going on here? Objectively, this seems really bad. ChatGPT seems to be unable to distinguish between what the user says and what documents or websites say the user said.
linkedin.com