Skip to content
Satire · Sourced

Exhibit 03

Six court cases that never existed

In 2023, a personal-injury case against the airline Avianca was working through a federal court in Manhattan. The plaintiff's lawyers filed a brief citing a string of supportive precedents: Varghese v. China Southern Airlines, Martinez v. Delta Air Lines, and several others. The opposing counsel could not find them. Neither could the judge. They did not exist.

What the machine did

A lawyer had used ChatGPT to do the legal research. The chatbot produced six cases that looked completely real: proper names, plausible docket numbers, internal quotations, and citations to the federal reporters. Every one was fabricated. When the lawyer grew suspicious and asked the chatbot whether the cases were genuine, it reassured him that they were and offered to provide full text. The full text was invented too.

"Six of the submitted cases appear to be bogus judicial decisions with bogus quotes and bogus internal citations."

Judge P. Kevin Castel, U.S. District Court, S.D.N.Y., Mata v. Avianca, 2023

What happened next

The judge sanctioned the two lawyers and their firm, ordering a 5,000 dollar penalty and requiring them to notify the real judges whose names had been falsely attached to the fake opinions. The story became a global cautionary tale. And then it kept happening. Courts in multiple countries have since caught filings built on AI-invented citations, from a Trump-adjacent legal team to attorneys for a major airline's own outside counsel. Researchers now track the growing list. The tool that was supposed to make legal work faster has generated a brand-new category of professional malpractice.

Why it belongs in the catalog

A fabricated citation is the hallucination problem with the stakes turned up. The output was fluent, formatted, and specific, which is exactly why it was dangerous. The model has no concept of a real case versus an invented one; both are just plausible sequences of legal-sounding text. The "ask it if it is sure" instinct fails completely, because the system will defend its fiction with the same confidence it used to create it.

Sources