Gujarat HC bars AI from judicial decision-making, drafting of orders and judgments

The Gujarat High Court has prohibited the use of Artificial Intelligence (AI) in judicial decision-making, reasoning and drafting of orders or judgments, according to a policy unveiled on 4 April.

The policy, released at a conference of district judiciary judges, states that AI tools cannot be used for “any aspect of judicial decision, adjudication, reasoning… or final judgment”.

Strict ban on adjudicatory use

The court has barred AI from being used in:

• Judicial reasoning and interpretation of law

• Drafting of orders, judgments or bail decisions

• Determination of rights, liabilities or sentencing

• Evaluation of evidence or facts

It also prohibits the use of AI for generating or altering evidence, or relying on AI-generated citations and legal references without independent verification.

“AI shall not be used… for any substantive adjudicatory process,” the policy states.

Concerns over risks

The High Court flagged several risks associated with AI tools, including “hallucinations, bias, confidentiality breaches, and erosion of judicial independence”.

It stressed that such risks require “institutional discipline” and careful handling.

The policy also bars entering sensitive or confidential information — including details of parties, witnesses or pending cases — into AI systems.

The court underscored that judges remain fully responsible for their decisions.

“A judge is personally responsible for every order, judgment, and observation… and this cannot be delegated,” the policy states.

Similarly, court staff are responsible for verifying any AI-generated content used in official work.

Limited use permitted

While restricting AI in core judicial functions, the policy allows its use in a limited, support role, including:

• Legal research and identification of precedents

• Administrative tasks and case management

• Drafting internal documents, circulars and notices

• IT-related automation and training materials

However, all AI-generated outputs must be reviewed and verified by a human officer before use.

‘Support tool, not substitute’

The court said AI should function only as a “decision-support and administrative efficiency tool” and not replace human judicial reasoning.

By limiting AI’s role to “the narrowest conceivable scope”, the policy aims to preserve judicial independence while improving efficiency and access to justice.

The move places clear boundaries on the integration of emerging technologies within the judicial system while reaffirming human primacy in legal decision-making.