Use of Generative AI in Court Filings
A Court of Appeals held that AI use alone does not invalidate an appeal and emphasized lawyers’ responsibility for their contents.
Chamber I of the Court of Appeals and Guarantees in Criminal Matters of the City of Moron examined the use of generative artificial intelligence in an appeal the defense counsel had filed for one of the defendants and held that its use is not, in itself, grounds for procedural nullity.
The issue arose after the lawyer included in the appeal excerpts from a conversation with an AI system, including instructions to reformulate grounds of appeal, suggestions for organizing arguments, and part of a response generated by the tool. The Court found that these elements objectively established the use of artificial intelligence and showed that part of the interaction had been incorporated into the filing “without sufficient professional review and editing”.
However, the available evidence did not establish which AI tool had been used, what proportion of the filing had been generated with it, whether there had been an intent to mislead, whether the defense strategy had been entirely delegated to the tool, or whether all statements and citations in the appeal were false. The Court thus distinguished using these tools from adopting their outputs without proper review. It explained that the problem is not using AI, which is not generally prohibited, but adopting AI outputs without the required verification, validation, and review.
The Court also examined the provisions from the Province of Buenos Aires Code of Criminal Procedure that govern the admissibility of appeals and procedural nullities. It noted that articles 433 and 442 require specific grounds of appeal and supporting reasoning but held that compliance with these requirements must be assessed based on the filing’s substance and its capacity to challenge the decision under review, rather than the method used to prepare it. Accordingly, using AI is not by itself grounds for nullity, which would require a legal basis and showing specific harm.
As to quotes from precedents, the Court highlighted that a reference that cannot be identified or verified cannot be relied upon as authority to support an argument, although that irregularity does not warrant presuming that the entire appeal is fictitious. Drawing on a prior decision, it noted that any dismissal of an appeal for insufficient grounds must be based on the inadequacy of those grounds, rather than on the mere suspicion that artificial intelligence was used.
This ruling clarifies the procedural implications of using generative AI in the legal practice. While these tools may assist in drafting court filings, their use does not relieve lawyers of responsibility for substantiating their arguments or verifying the sources they quote.
This insight is a brief comment on legal news in Argentina; it does not purport to be an exhaustive analysis or to provide legal advice.