Lawyers cited fabricated cases ChatGPT invented

What happened

Lawyers representing Roberto Mata used ChatGPT to research a legal filing. The model invented cases and citations that didn't exist - and they made it into the brief filed with the court.

The court found the cited decisions were fictional. The attorneys were sanctioned $5,000 and required to notify their client and the judges whose names had been fabricated into the fake citations.

The business problem

Unverified research, no review step, reputational exposure, legal risk, and a client relationship damaged by it.

What could have helped

• Make source verification mandatory, not optional
• Use an AI-output verification checklist for anything client-facing
• Train staff on where AI-generated "facts" can be wrong
• Restrict AI in high-risk tasks until a review process exists

Where this points