Improper Use of Artificial Intelligence in Court Documents: Lawyers’ Professional Liability

In recent months, Italian court case-law has been adopting an increasingly restrictive approach on the improper use of Artificial Intelligence (AI) in the drafting of court documents. Several decisions have sanctioned, in the context of judicial proceedings, the automatic and unverified use of AI and the uncritical reliance on AI tools, with significant implications for lawyers’ professional liability.

  • Court of Florence (order, 14 March 2025): although no sanction was imposed under Article 96 of the Italian Code of Civil Procedure, the Court clarified that lawyers shall always verify the accuracy of cited sources, even when using tools such as ChatGPT—highlighting instances in which references to court case law had been fabricated or misquoted.
  • Court of Turin (judgment no. 2120/2025): in this case, the Court imposed sanctions for abuse of judicial process: the judicial deed —clearly AI-generated— included abstract, disordered, and irrelevant references, with no meaningful connection to the specific dispute.
  • Court of Latina (judgment no. 1034/2025): a particularly severe sanction was imposed for “cut and paste” pleadings, attributable to the large-scale use of AI, containing arguments that were manifestly unfounded and unrelated to the issues to be decided by the court (i.e.,thema decidendum“).
  • Italian Supreme Court (order no. 34481/2025): the Supreme Court also warns judges against the risk of “abdicating” their evaluative function, stressing that AI can never replace the human critical assessment that is essential to ensuring judicial independence, neutrality and impartiality.

The conclusions from the court case-law is clear: AI is a legitimate and valuable tool, but it cannot be used mechanically or without control. In particular, what courts criticize is not the use of technology as such, but the professional’s failure to exercise independent critical judgment. A lawyer who relies on AI without proper verification exposes himself (and his client) to procedural sanctions, professional liability, potential damages claims. AI may support legal work, but it cannot replace legal expertise, professional experience and accountability.