The Gujarat High Court on Saturday issued a policy regulating the use of artificial intelligence in judicial administration and courts, stating that artificial intelligence cannot be used in adjudication processes, including making judicial decisions, drafting judgments or orders and evaluating evidence, and holding judges personally liable for all orders and judgments passed in their names.
The policy, titled “Policy on the Use of Artificial Intelligence in Judicial Administration and Courts,” states that while AI can improve efficiency, research and administrative performance, the core function of adjudication in the judiciary must remain in the hands of judges.
“The essence of making judgments, weighing evidence, interpreting the law, applying legal principles to facts, exercising discretion, and making reasonable decisions belong exclusively to the domain of human reason,” the report said.
The policy prohibits the use of artificial intelligence in making judicial decisions, judicial reasoning, drafting an order or judgment, bail or judgment considerations, interpreting facts, weighing arguments, determining rights and obligations, interim orders or final rulings.
In addition, the policy strictly prohibited the entry of confidential court data into public AI tools, including names and addresses of parties or witnesses, details of pending proceedings, privileged communications, sensitive personal data such as health or financial information and any evidence or documents submitted in a case. It is also prohibited to use artificial intelligence to fabricate or alter evidence or to rely on artificial intelligence-generated citations without independent verification from reliable legal sources.
This policy applies to the entire judicial and administrative framework under the High Court of Gujarat, including the Registry of the High Court and District Judiciary, covering all subordinate courts under its jurisdiction. It applies to judicial officers, court staff, paralegals, trainees, interns, paralegal volunteers, ministerial, administrative and technical staff and covers all court-related work, whether carried out within the court premises or remotely.
This comes months after the Supreme Court set aside a lower court order that cited fake rulings generated using artificial intelligence, warning that decisions based on “non-existent” and “spurious” rulings may amount to misconduct. “At the outset, we must declare that the decision based on such allegedly non-existent and spurious provisions does not amount to error in decision making,” the Supreme Court said in its February 27 order.
The Supreme Court said the policy was necessary because of the rapid growth of artificial intelligence tools and the risks associated with them, including hallucinations, bias, breaches of confidentiality and erosion of judicial independence. She said that artificial intelligence should only be used as a tool to support decision-making and administrative efficiency, and not as a substitute for judicial reasoning.
“Each judge remains fully and personally responsible for every order, judgment and notice issued in his or her name,” the policy reads.
This policy is issued under Articles 225 and 227 of the Constitution of India which deals with the jurisdiction of the Supreme Court and supervisory powers over subordinate courts and is to be read in conjunction with the Information Technology Act, 2000, Protection of Digital Personal Data Act, 2023, Contempt of Courts Act, 1971 and Gujarat High Court Rules, 1993.
However, the Supreme Court has allowed limited use of AI for administrative and research purposes. These include legal research support such as judgment retrieval and precedent setting, improving the language and structure of drafts where the legal reasoning remains that of the judge, verified machine-assisted translation, hearing transcription, case management, scheduling, cause list management, statistical reporting and administrative tasks. AI can also be used to generate initial lists of relevant cases or laws for further research, but this cannot replace standard legal databases such as SCC Online or AIR, the policy said.
The policy stipulates that all AI-generated case citations, legal references and summaries must be independently verified and compared to the original rulings or statutes before use, and machine translations and AI transcriptions must also be verified before being relied upon in court proceedings. It further stipulates that the use of artificial intelligence will not serve as a defense in cases of error, misconduct or negligence.
Violation of any provision of the Policy will be treated as misconduct and may result in administrative or disciplinary action, notwithstanding any civil or criminal liability under applicable laws. The Supreme Court said that the policy will remain in effect until it is revised and can be amended as per the directions issued by the Supreme Court or future policies on the use of artificial intelligence in courts.
