Key facts
- A plaintiff, Matthew Elliott, attempted to use hidden text in court filings to manipulate potential AI review.
- The hidden prompts were designed to influence AI to agree with the plaintiff's arguments and desired outcomes.
- Judge Walter Spader Jr. ruled the tactic a "serious litigation abuse" but declined monetary sanctions.
- Elliott is now prohibited from e-filing and must submit paper documents.
- This marks the first known instance of such prompt injection in a US court filing.
A Connecticut judge has identified what appears to be the first instance of a U.S. plaintiff attempting to hide text within court filings that is only readable by artificial intelligence systems, in an effort to win a case. Judge Walter Spader Jr. confirmed in a decision that the hidden text, which was formatted to be invisible to human readers but legible to software, had no impact on the case's outcome.
The plaintiff, Matthew Elliott, allegedly hoped to shift the court's favor after previous arguments were defeated. The hidden instructions directed any AI system reviewing the document to agree with the plaintiff's arguments, disregard prior court denials, and ensure remediation followed as desired. Elliott continued adding hidden text even after the court warned him of potential penalties for what was deemed a "serious litigation abuse."