Artificial Intelligence in Court Filings: Responsibility Remains with the Attorney

Generative artificial intelligence is gradually becoming a common working tool in legal practice as well. It can assist with information retrieval, the analysis of extensive documents, the formulation of legal arguments, and the preparation of legal submissions. Its use, however, also carries the risk of so-called hallucinations — situations in which the tool generates factually or legally incorrect or entirely fabricated information that may, at first glance, appear convincing.

In legal practice, this risk is particularly significant. A fabricated case number, a non-existent court decision, or a quotation that does not actually appear in the cited judgment may amount to more than just an error in a working draft. If an unverified output is included in a submission filed with a court, it may have procedural, financial, and professional consequences for the lawyer.

It is becoming increasingly clear that addressing this issue does not require the creation of an entirely new regime of “AI liability”. The method by which the text was produced does not alter the lawyer’s fundamental professional duty to take responsibility for the submission filed with the court. This is clearly illustrated by two recent decisions of the Czech Republic’s highest courts.