Austrian Supreme Court gives AI-generated appeal short shrift

Austrian Supreme Court gives AI-generated appeal short shrift

Austrian Supreme Court gives AI-generated appeal short shrift

The first AI-generated, hallucination-riddled appeal has reached the Austrian Supreme Court – and the court’s response was extraordinary (14 Os 95/25i).

Courts in the U.S., Canada, New Zealand, and Australia have already confronted AI-generated fake citations and fabricated authorities in legal filings. Some imposed sanctions. Others warned of contempt or disciplinary consequences.

Austria’s highest court has joined the list.

In a remarkable judicial dressing-down, the court effectively says that the filing was so corrupted by AI hallucinations that it failed at the threshhold level of legal argumentation:

“The remainder of the submission, which is riddled with numerous misquotations (concerning, on the one hand, alleged procedural outcomes and, on the other hand, Supreme Court decisions that, for the most part, do not exist at all or, in any case, do not pertain to the stated subject matter) and was apparently prepared without professional review […] by so-called “artificial intelligence” does not meet the requirement to clearly and specifically identify grounds for nullity, i.e., to present facts justifying nullity at a level of argumentation appropriate for the Supreme Court as the highest court […] and therefore does not even come close to meeting this requirement and thus precludes a substantive response.”

Most strikingly, the court did not merely criticize the filing, it effectively refused substantive engagement with it because the hallucinations undermined the appeal at a threshold level.

Notably, the court explicitly referenced Sec. 9 of the Austrian Bar Rules and disciplinary case law pursuant to which even negligent inaccuracies can damage the honor and reputation of the legal profession and lead to disciplinary sanctions. This could be read as a pointed call to the bar to take disciplinary action against such practices.

The principle crystallising from Vienna to New York is clear. AI is not the problem – uncritical reliance on AI is. One can only hope that the lesson has now been learnt, and that the courts will be spared any further encounters with phantom case law or hallucinated procedural outcomes.

 

Author: Katrin Hanschitz

Related Insights and Influence

Diversity