Brazilian judge publishes error-strewn AI-generated decision

Occurred: 2023

A Brazilian federal judge who used ChatGPT to write an error-strewn legal decision is being investigated by Brazilian authorities. 

Brazil's National Justice Council (CNJ) said it had summoned Judge Jefferson Rodrigues to explain why he had published a decision strewn with legal errors generated by the ChatGPT chatbot

In his ruling, Rodrigues had included incorrect details on previous court cases and legal precedent, wrongly attributing past decisions to the country's Superior Court of Justice.

The judge stated that the ruling was written by a 'trusted advisor,' with help from AI, and called the situation a 'mere mistake,' blaming 'the work overload facing judges.'

System 🤖

Operator: Jefferson Rodrigues
Developer: OpenAI
Country: Brazil
Sector: Govt - justice
Purpose: Generate legal decision
Technology: Chatbot; NLP/text analysis; Neural network; Deep learning; Machine learning; Reinforcement learning
Issue: Accuracy/reliability; Accountability
Transparency: Governance; Marketing