tech
Suspecting court of using AI, man injected prompts in filings to try to win case
Judge warns pro se litigants are using chatbots wrong and getting desperate.
TL;DR
- A judge has identified what appears to be the first US case of a plaintiff using hidden, AI-readable text in court filings.
- The plaintiff, Matthew Elliott, attempted to use prompt injection to manipulate AI systems into favoring his arguments.
- The hidden text was formatted to be invisible to humans but legible to AI, aiming to influence the court's reading of the filings.
- The tactic failed, and the court ruled on the merits of the case without being influenced by the hidden prompts.
- Elliott faced sanctions for "serious litigation abuse," including a prohibition from e-filing.
- The judge warned that this tactic, while currently ineffective, sets a dangerous precedent and may become more common.
- The article notes a similar case in Brazil where attorneys faced monetary sanctions for using prompt injection.