Incident file · Knowledge
Mata v. Avianca (2023)
court sanctions, $5,000 fine against attorneys and firm
The system
ChatGPT, used directly by attorney Steven Schwartz as a research tool while preparing a personal injury brief against Avianca airlines. This was a plain generative chatbot: no retrieval grounding, no citation verification, no connection to a real legal database.
What was observed
The brief Schwartz filed cited six court decisions supporting his argument: Varghese v. China Southern Airlines, Martinez v. Delta, Shaboon v. Egyptair, Petersen v. Iran Air, Estate of Durden v. KLM, and Miller v. United. Opposing counsel and the court could not find any of them.
Stop and think
- Schwartz asked ChatGPT itself whether the cases were real, and it said yes. What does that tell you about whether a generative model can be trusted to verify its own output?
- What single step, inserted anywhere in this workflow, would have caught the problem before the brief was filed?
- Is this a "RAG failure" or something else entirely? What would have to be different about the system for "RAG failure" to even apply?
Answer the questions above before you read on: what failed, why, and what would have stopped it.