Gujarat High Court bars AI use in decision-making, judgment drafting

Ahmedabad: The Gujarat High Court has prohibited the use of Artificial Intelligence for any form of decision-making, judicial reasoning, order drafting or judgment preparation, bail sentencing considerations, or any substantive adjudicatory process.

According to the high court’s AI policy, unveiled on Saturday at a conference of district judiciary judges in Gujarat, AI should be used to improve the speed and quality of justice delivery, rather than as a replacement for judicial reasoning.

As per the policy, these technologies “carry substantial risks — including hallucinations, bias, confidentiality breaches, and erosion of judicial independence — that must be managed with care and institutional discipline”.

It says that by confining AI to the narrowest conceivable role, purely anonymised, metadata-driven case allocation and research of legal principles, “human supremacy in justice delivery is reaffirmed and limited technological assistance is harnessed to reduce administrative imbalances that might otherwise delay access to justice”.

“Broadly, Artificial Intelligence shall not be used—directly or indirectly—for any aspect of judicial decision, adjudication, reasoning, application of law, interpretation of facts, weighing of arguments, determination of rights/liabilities, sentencing, bail, interim orders, or final judgment,” as per the policy document.

Also, AI shall not be used to find facts, law, or operative order in any judicial proceeding, nor shall it be used for sorting, classification of evidence, organisation of evidentiary material, or any task involving evaluation or categorisation of proof, it says.

AI cannot be used to author, generate, or substantially compose any judgment, final order, or binding legal ruling, even if subsequently reviewed by a judge, according to the policy document.

One cannot enter names, addresses, or identifying information of parties, witnesses, or advocates, details of pending proceedings or unreported orders, privileged communications or confidential legal strategies, and sensitive personal data, it says.

AI shall not be used to generate, fabricate, embellish, or alter evidence in any form, using AI-generated citations, case references, or statutory provisions without independent verification from authoritative primary sources is also prohibited, as is drafting, correcting or summarising any office note or submission, it says A judge is personally responsible for every order, judgment, and observation issued under their name, and this cannot be delegated, shared with, or diminished by the use of any AI tool, it says.

Every court officer is personally responsible for the accuracy and appropriateness of any AI-generated content used in the performance of their official duties, stresses the policy “AI in judiciary should be designed as a decision-support and administrative efficiency tool, not as a replacement for judicial reasoning. With proper safeguards—such as transparency, human supervision, and protection of confidential information—AI can significantly strengthen case management and improve the speed and quality of justice delivery,” it states.

The policy requires a qualified human officer to review, verify, and be responsible for any AI-generated output before it is acted upon, filed, published, or communicated. AI-generated content, including case citations and statutory references, must be independently verified against authoritative primary sources before use, it says.

At the same time, the policy allows judicial officers and court staff to leverage AI tools to improve productivity, reduce administrative burden, and enhance access to justice, while preserving judicial independence and the sanctity of judicial decision-making.

The policy allows AI tools for various administrative and productivity tasks, code generation or automation for IT department tasks, creating presentations or templates for internal training purposes, and drafting and improving circulars and notices, information about which is in the public domain.

AI is also permitted to be used for legal research, retrieval or analysis of judgments, identification of precedents, statutory interpretation, or any preparatory intellectual work supporting adjudication, “but with all human conscience and subject to verification by applying mind”.