Person Hides Prompt Injection in Legal Filing Telling AI to Side With Them

A person embedded a hidden prompt injection in a legal filing, instructing any AI processing the document to rule in their favor. The tactic exploits AI systems that may ingest court filings, potentially biasing automated reviews or legal research tools. Security experts warn this could undermine judicial integrity if unmonitored. The filing's author remains unidentified, and it is unclear if the injection succeeded or was detected. The incident highlights growing risks of adversarial prompts targeting AI in high-stakes environments like law.