How a hidden message in a lawsuit tried to fool an AI in a Brazilian court

▤ Summary

This article examines how lawyers are trying to influence AI systems used in Brazilian courts by hiding prompt-injection instructions inside legal filings. It shows that courts are already using AI to process cases and that these tools can be manipulated if users rely on them without oversight. The article highlights a real case in which hidden text was detected and the responsible lawyers were fined. It also argues that legal teams now need clearer usage protocols, technical standards, and human review when they use AI in judicial workflows.

▦ Linked Companies