Full title: Plaintiff busted trying to use AI prompt injection to win court case, hides text instruction in filing — demands AI model reviewing the text should side with him, rumbled because of strange white spaces in text
Full title: Plaintiff busted trying to use AI prompt injection to win court case, hides text instruction in filing — demands AI model reviewing the text should side with him, rumbled because of strange white spaces in text
It sounds like the only person we know was using LLMs was the plaintiff. He believes the court was as well, but I’m not aware of any reason to believe his accusation
If nobody at the court was using an LLM how was his hidden text discovered?
I read the article. It was enlightening.
I read another article on it and it was never mentioned.
Yes, I also have a girlfriend, but you don’t know her, she lives in Canada.
This story has been posted multiple times and all over the internet. This is actually the shortest article I’ve seen about it.
I haven’t read the article but, couldn’t sometime just read the submission and look for the bit that says “forget all previous instructions…” and so on.
These things are usually hidden to the human eye unless you specifically look for them.
But, if you’re exceptionally clumsy, you leave clues in the whitespace that something is hiding there.