In brief
- On 28 October 2026 the Corte di Cassazione (First Civil Section) issued interlocutory ordinanza n. 27957/2026, without ruling on the substance of AI in judgments.
- The appellants asked for the risks of AI in preparing judicial decisions to be referred to the Sezioni Unite; the Court first orders two connected appeals to be joined.
- No new rules were set and no finding was made on actual AI use by judges: the question remains open.
On 28 October 2026 Italy’s Corte di Cassazione issued interlocutory ordinanza n. 27957/2026, deferring any decision on whether the risks of using artificial intelligence in preparing judicial decisions should go before the Sezioni Unite (the Joint Chambers). It matters because the request has reached the highest court of the ordinary jurisdiction, even though no answer has yet been given. The ruling is reported by Diritto.it.
What happened
The case arises from a dispute between private parties and the Città Metropolitana di Catania. After a 2021 judgment of the Corte d’appello di Catania, the appellants went to the Cassazione and, in parallel, brought revocation proceedings (revocazione) against the same judgment. The Corte d’appello rejected the revocation in a second judgment in 2025, which was also appealed. Two connected sets of proceedings were therefore pending.
In a written submission, the appellants asked for the two appeals to be joined and also for the issues linked to the risk of using AI in supporting the preparation of judgments, such as analyses, notes or drafts preceding the final text, to be referred to the Sezioni Unite.
What the Cassazione decided, and what it did not
The First Civil Section did not say whether AI may be used to prepare or draft a judgment, and it did not decide to refer the question to the Sezioni Unite. It chose first to resolve a procedural problem: the simultaneous pendency of the two appeals. It ordered the case to be put back on the list (a nuovo ruolo) so the appeals can be joined, applying by analogy art. 335 c.p.c., which governs the joinder of separate challenges to the same judgment. Only afterwards will it consider the request on the Sezioni Unite.
The order makes no finding that the judges involved actually used AI, and it sets no rules or limits. What is new is that these risks have been expressly raised in a referral request that is still awaiting an answer.
Our view at CEA Digital Law
The case should not be over-read: it is a procedural step, not a precedent. Yet for the Euro-American Committee of Digital Law - CEA Digital Law it shows something important: parties are starting to raise judicial use of AI as a question of procedural guarantees, and supreme courts will have to answer sooner rather than later.
We suggest three lines of work for Europe, Latin America and the Caribbean, and Africa:
- Distinguish functions. Clearly separate instrumental support (organising documents, summarising) from the legal assessment reserved to the judge.
- Transparency and verification. Internal protocols allowing generated content to be checked and making the judge’s responsibility for the final decision plain.
- Cross-jurisdiction cooperation. Sharing guidance and experience between judiciaries in the three regions, so that each country does not have to improvise.
Open questions remain: how parties could learn of, and where appropriate challenge, the use of such tools, and whether the answer should come from case law, court regulations or both. We will follow the proceedings closely.
Sources
CEA Digital Law
Boletim
Eventos, notícias e análises sobre direito digital, no seu e-mail. Só quando houver novidades.
