Magid Nauef Láuar, of , to acquit a 35-year-old adult of vulnerable rape against a 12-year-old child, contravenes the (National Council of Justice) rule for the use of , concludes a technical opinion from -PR (Brazilian Bar Association Section of ).
According to the entity, the decision, which contains fragments of commands for generative AI, is at odds with regulations, especially regarding the need for human review.
When contacted, the TJ-MG (Court of Justice of the State of Minas Gerais) said that the magistrate will not comment.
As shown, the full ruling contains a prompt that says “Now improve the exposition and reasoning of this paragraph”, followed by an excerpt. Below appears an AI-summarized version of the same content.
“Such a finding shows, at the very least, that the official text of the ruling did not undergo human verification, review and supervision after the text was generated by AI”, states a note from OAB-PR.
“The leaked prompt added to two repeated paragraphs is evidence that this supervision did not occur, at least in an effective or sufficiently safe manner. This state of affairs triggers the absence of the rigor that is due in the justification of judicial decisions.”
The note is signed by lawyer Eduardo Caldi, president of the entity’s Artificial Intelligence Commission. For him, the violation may eventually give rise to punishment, but it depends on administrative procedures within the jurisdiction of the CNJ.
“The fact is that it is every citizen’s right to have their case judged by a magistrate, a human person. If the decision is made by a computer system, an artificial intelligence, the CNJ’s own resolution provides for measures and resources”, he tells the Panel. “The rule is clear.”
To debate the responsible use of AI in the Judiciary, the OAB sections in Paraná, Rio Grande do Sul and Santa Catarina will promote the 1st South-Brazilian Prerogatives Meeting, on March 19th and 20th, in Foz do Iguaçu (PR).
Approved just over a year ago, the CNJ resolution that regulates the use of artificial intelligence in the Judiciary requires human participation and review at all stages of use.
The rule establishes that AI has an auxiliary and complementary nature and prohibits its use as an “autonomous instrument for making judicial decisions without due guidance, interpretation, verification and review by the judge, who will remain fully responsible for the decisions taken and the information contained therein”.
The TJ-MG states that the use of artificial intelligence tools in court offices meets regulatory guidelines and that, internally, it has promoted educational and informative actions on the topic, such as courses, podcasts and training.
The court says that these are initiatives with guidance on the functioning of these resources and their ethical use, highlighting the need for human supervision. They are aimed at both magistrates and advisors and civil servants.
The trial on the rape case took place last month, and was carried out by the judge himself after negative repercussions of the acquittal. Today after being the target of reports of alleged sexual abuse.
LINK PRESENT: Did you like this text? Subscribers can access seven free accesses from any link per day. Just click the blue F below.
