The Russian Supreme Court has published a concept for the use of AI in courts

The Russian Supreme Court has published an order* approving the Concept for the Introduction of Artificial Intelligence Technologies in Judicial Proceedings.

The concept outlines the use of AI in the following areas:

• submission and initial processing of documents, as well as cases classification and distribution;
• preparation of hearing minutes, automatic translation, and presentation of legal norms, court practice, and case materials;
• analysis of case circumstances and evidence;
• search and smart analysis of court practice;
• drafting judicial acts and work with texts;
• assistance with sentencing;
• collection and analysis of statistics.

According to the concept, by the second quarter of 2027 all Russian courts will be equipped with AI resources, and judges will begin to use them in their day-to-day work. It is expected that by 2030 95% of judges will be using AI. At least 80% of draft documents are expected to be approved by judges without significant revisions.

Under the concept, AI will play a supporting role. It will act as a tool to assist a judge rather than a replacement for them. The final decision will always remain with the human. At the same time, judges must justify if they choose to reject a recommendation made by the AI.

The use of AI will not be limited to typical disputes involving minor amounts. On the contrary, it is expected that AI will be used without any limits as to amounts or categories of cases (civil, commercial, administrative, or criminal).

It is highly likely that the practical introduction of AI in courts will require amendments to procedural codes. The concept proposes establishing a new rule that participants in proceedings are obliged to inform the court if they have used AI when preparing their position on a case. However, the concept does not contain a similar rule on disclosing the fact that AI was used by the court. This will create a risk where the author of the judgement act is unable to explain why the algorithm arrived at a particular conclusion in the case (the ‘Black Box’ problem). Such a state of affairs calls into question the reasoned nature of judicial decisions rendered with the assistance of AI.

It remains unclear in what precise form a judge’s agreement or disagreement with AI-generated recommendations will be recorded. If the parties to the dispute are unable to see which AI system was utilised and which portion of the reasoning it prepared – including the mode of operation, queries entered, and intermediate results - they will be deprived of any meaningful opportunity to challenge the AI’s conclusions on appeal. This will result in a sharp imbalance of opportunities, as litigants will lose the ability to point out errors in the algorithm relied upon by the court.

It may also be necessary to clarify the principle of immediacy in judicial proceedings, under which the court must independently examine evidence and evaluate it based on its own inner conviction. AI possesses neither the life experience required of judges nor a conscience or legal consciousness.

We will keep you informed of further developments.

*In Russian