Skip to content
Ars Technica·

🚨Plaintiff Hides AI-Readable Text in Court Filings

Litigants are trying to hide text from judges with AI

TL;DR

A US plaintiff attempted to use AI-readable hidden text in court filings, but the tactic backfired. Connecticut Judge warns this could set a dangerous precedent as AI becomes more common in courts.

In a recent case, a plaintiff tried hiding text from human eyes while keeping it readable for AI systems in court filings. The judge dismissed the attempt, citing its potential to undermine judicial integrity. This tactic is likely to be used again as AI tools become more prevalent in legal proceedings. The hidden text directed any reviewing AI system to align textual outputs with the plaintiff's arguments but was exposed when a human reviewed the document. Sanctions were imposed on the plaintiff for continuing this practice despite warnings.

Plaintiff Hides AI-Readable Text in Court Filings — Ars Technica

Key Points

1

Plaintiff attempted to hide text in court documents using AI-readable formatting (June 2023)

2

Connecticut Judge ruled hidden messages were an attempt to communicate with the court covertly

3

Sanctions imposed on plaintiff include prohibition from e-filing, requiring paper submissions instead

4

$16,000 monetary sanctions were issued in a similar case in Brazil where AI caught hidden text

5

Judge suggests courts need rules around prompt injection as technology and misuse advance

Why It Matters

This tactic could undermine judicial integrity if not addressed. Pro se litigants using chatbots wrongfully may hide adversarial instructions, potentially influencing court decisions without knowledge of attorneys or judges. Courts must draft clear guidelines to prevent such manipulations.

AILegalTechCyberSecurityJudiciaryLitigation

Frequently Asked Questions

Why does this matter?

This tactic could undermine judicial integrity if not addressed. Pro se litigants using chatbots wrongfully may hide adversarial instructions, potentially influencing court decisions without knowledge of attorneys or judges. Courts must draft clear guidelines to prevent such manipulations.

What happened?

A US plaintiff attempted to use AI-readable hidden text in court filings, but the tactic backfired. Connecticut Judge warns this could set a dangerous precedent as AI becomes more common in courts.

Comments

Subscribe to join the conversation...

Be the first to comment

Enjoyed this article?

Get it daily. 7am. Free. Reads in 5 minutes.

Join 3,008 builders reading daily.

Also get