First Known Hidden AI Directive in Court Filing Raises ‘Massive’ Concern
A judge said a Connecticut plaintiff hid messages in court filings instructing any AI reviewing them to side with him.
A Connecticut judge has expressed concern over a hidden AI directive found in a court filing, raising questions about the potential misuse of artificial intelligence in legal proceedings. Matthew Elliott, representing himself in a lawsuit against New York Bariatric Group, attempted to instruct any artificial intelligence reviewing the documents to side with him.
By embedding tiny, nearly invisible text within the filing, Elliott aimed to direct the AI system to produce output favoring his position and to treat a prior ruling as an error. The judge, Walter Spader Jr., likened this technique to a "prompt injection" and stated that he was unaware of any prior U.S. decisions addressing the issue.
Spader called the messages "effectively invisible" to a human reader but "ordinary, machine-readable text" to any system processing the file. Elliott claimed the hidden directives were designed as an "audit" to determine whether the court was using AI to process his filing. However, the court discovered the messages after noticing extra white space within the pleadings.
Elliott's use of hidden AI directives was sanctioned by the judge, who revoked his electronic filing privileges. While AI tools can be beneficial in preparing legal filings, the judge warned that their misuse could undermine public trust in judicial institutions.
Written by urgent.news from Newsweek's reporting — not their text. Machine-written — it may contain errors, so check the original before relying on it.
