Connecticut man sanctioned for hiding AI prompt injections in legal filings
A self-represented litigant in Connecticut filed court documents containing hidden instructions designed to manipulate artificial intelligence systems. Matthew Elliott embedded white text in a 3-point font throughout his July 2026 filings in a case against the New York Bariatric Group, instructing any AI reviewing the documents to side with him and overturn a prior ruling. A court staff member discovered the hidden text by noticing unusual white space on paper printouts. Judge Walter Spader Jr. issued a decision finding this the first documented prompt injection attack on a U.S. court and the first judicial sanction for the practice. Elliott claimed he was conducting a system audit, but continued hiding text even after receiving a warning, including jokes and SpongeBob references. The judge barred Elliott from electronic filing, requiring all future papers be submitted in person. Legal experts warn the practice poses serious risks to court integrity.
Key facts
- Matthew Elliott hid white 3-point font text in court filings on July 24, 2026, directing AI systems to side with his position
- Court staff discovered the hidden injections by noticing unusual white space on paper printouts of the filings
- This is the first documented prompt injection attack on a U.S. court and the first judicial sanction for the practice, according to legal analysis
- Judge Walter Spader Jr. sanctioned Elliott on August 6, 2026, by revoking his electronic filing privileges and requiring in-person paper submissions
- Elliott continued hiding text in subsequent filings even after receiving a July 31 warning order from the court
- Elliott claimed the filing was a system audit, but the court found evidence of intentional manipulation including jokes hidden in later filings
How different outlets framed it
Left-leaning coverage: Left-leaning tech outlets like Gizmodo and Techdirt frame this as an example of litigation abuse and potential threats to court systems. They emphasize the novelty and seriousness of prompt injection attacks, and stress the need for safeguards as AI becomes more integrated into legal proceedings.
Limits of this briefing
The story is emerging and being covered primarily by tech and legal specialty media rather than mainstream news outlets, so the full political framing may not yet be fully developed. Right-leaning outlet commentary was not prominently featured in available sources.
Sources
- 404 Media: Person Hides Prompt Injection in Legal Filing Telling AI to Side With Them Centre
- Gizmodo: Dude Reportedly Hides Prompt Injections in Legal Filing, Just in Case Judge Is Really That Lazy Left
- Techdirt: Pro Se Plaintiff Caught Hiding Prompt Injections In Court Filings; Responds By Hiding More Left
- Reason (Volokh): Court Faults Self-Represented Plaintiff for Including Hidden "Prompt Injection" in Court Filing Centre
- Harris Beach Murtha (Law Firm): The First Documented Prompt Injection Attack Aimed at a U.S. Court Centre