No AI for judgments: Gujarat High Court sets strict rules for courts

The Gujarat High Court has prohibited the use of Artificial Intelligence (AI) in judicial decision making, judgment drafting, and legal reasoning , drawing a clear line between technological assistance and the role of judges in courts.

The new AI policy , unveiled on Saturday during a conference of district judiciary judges in the state, states that AI should only be used to improve efficiency in the justice system and not as a substitute for human judgment .

According to the policy, AI cannot be used for any form of adjudication , including interpretation of facts, application of law, bail considerations, sentencing decisions, or preparation of interim and final orders. The High Court emphasised that judicial reasoning must remain entirely human driven .

The document also warns that AI technologies come with significant risks such as misinformation or “hallucinations”, algorithmic bias, confidentiality breaches, and threats to judicial independence . Because of these risks, the court has restricted AI to a very limited role within the judicial system.

Under the guidelines, AI cannot be used to find facts, evaluate or categorise evidence, organise evidentiary material, or generate judgments and binding legal rulings , even if such content is later reviewed by a judge.

The policy further bars court officials from entering sensitive information into AI systems, including details of parties, witnesses, advocates, pending proceedings, confidential communications, or legal strategies.

Importantly, the High Court clarified that every judge remains personally responsible for the orders and judgments issued under their name , and this responsibility cannot be shared or delegated to any AI tool.

However, the policy allows limited use of AI for administrative and productivity related tasks . Courts may use AI for legal research, retrieving judgments, identifying precedents, preparing training materials, automating certain IT functions, and drafting internal circulars or notices that rely on publicly available information.

The High Court also stated that any AI generated output must be carefully reviewed and verified by a qualified human officer before it is used in official work. Legal citations and references produced by AI must be cross checked with authoritative sources.

The policy emphasises that AI should function only as a decision support and efficiency tool , helping courts manage cases faster while ensuring that the core responsibility of delivering justice remains firmly with human judges.