In continuation of our discussions on the proposed regulations for use of AI in judiciary released by Supreme Court for public Comments, we shall today look at Chapter III of the regulations related to Permissible and Prohibited uses.
Under this chapter there is a clear definition of what kind of AI is permissible and what is not. While DGPSI-AI and other frameworks rely on a Risk Assessment and classification of Risk as significant or not, since this regulation is sector specific, the regulatory draft directly defines what is permitted and what is not.
The following is the text of the regulation.
CHAPTER III PERMISSIBLE AND PROHIBITED USES
- Appropriate Authority to identify use of AI in Courts.–– The Appropriate Authority shall determine, from time to time, the areas of Court processes in which AI Systems may be utilised, in accordance with the provisions of these regulations.
- Permissible uses of AI.–– (1) Subject to prior approval in writing by the Appropriate Authority and to the supervision and verification of officers nominated for that purpose, AI Systems may be used for the following purposes, which are illustrative and not exhaustive:–
(a) case management (including identification of defects in new filings), cause list preparation, hearing scheduling and docket prioritisation;
(b) automated transcription of court proceedings, subject to mandatory review and certification of accuracy by a Designated Officer;
(c) translation of judgments, orders, pleadings and other legal documents, subject to human verification of accuracy and fidelity to the original;
(d) legal research, precedent retrieval, citation verification and document summarisation
(e) administrative functions including case filing assistance, defect scrutiny, record management and judicial resource allocation;
(f) conversational AI Assistants and guided chatbots to assist litigants and other stakeholders in accessing Court services and understanding procedural requirements, subject to human oversight of their functioning;
(g) accessibility services including text-to-speech, speech-to-text, Braille translation and visual assistance tools, for persons with disabilities or language barriers;
(h) document authenticity verification and fraud detection in administrative processes, subject to mandatory human review of all outputs before any action is taken;
(i) anonymisation of judgments, orders and Court records for publication in the public domain;
(j) analytical tools for judicial administration, court performance assessment and backlog monitoring and management; and
(k) auto-generation of prescribed formats, notices and summons with metadata merge including automated preparation of administrative documents.
(2) Any use of AI not enumerated under sub-regulation (1), or not otherwise specifically approved, shall require the prior written approval of the Appropriate Authority, which shall record reasons for the grant or refusal of such approval.
- Prohibited uses of AI.–– (1) The following uses of AI are strictly prohibited in all Court processes. These prohibitions are absolute and non-derogable, and shall not be subject to relaxation or modification by any authority under these Regulations, including under the power conferred by Regulation 19(1)––
(a) no personal data of any person shall be used to train, test, or refine any AI System without the prior approval of the Appropriate Authority and, where applicable, in compliance with applicable data protection law;
(b) no judicial outcome (including any judgment, order, or finding of fact or law) shall be reached through Algorithmic Decision-Making alone or solely on the basis of AI-generated information, data, or analysis and the human judicial authority shall be the determinative authority in all adjudicative decisions;
(c) no AI System shall perform the function of adjudication or sentencing in any matter without mandatory Human-in-the-Loop and any output of an AI System in relation to adjudicative or sentencing questions shall be treated as advisory only and shall be subject to independent judicial evaluation;
(d) no AI System shall be used for Risk Scoring for any purpose in Court processes, including the assessment of flight risk, prediction of recidivism, evaluation of bail eligibility, or determination of the credibility of parties or witnesses;
(e) no undisclosed, opaque, or unexplainable AI System shall be used in any Court process that may materially affect the lawful rights or personal liberty of any party;
f) no AI System shall be used to predict, profile, or infer the future conduct or behaviour of parties, accused persons, witnesses, or legal representatives in any Court process;
(g) no AI System shall be used for the surveillance or continuous monitoring of judicial officers, advocates, litigants, or any other person within or in connection with Court premises or Court processes, except as may be specifically authorised by applicable law for the time being in force;
(h) no AI-generated output shall be submitted to a Court as an independent source of evidence without full and transparent disclosure of its AI-generated character; and
(i) no AI System shall be used in any manner that may compromise the confidentiality of judicial deliberations or the independence of the judicial decision-making process.
(2) The prohibitions referred to in sub- regulation (1) shall be absolute and shall not be subject to relaxation or modification by any authority.
- Remedial measures by AI Committee.–– Every violation of any prohibition specified in regulation 20 shall be reported forthwith to the AI Secretariat, which shall be placed before the AI Committee, and the AI Committee shall, after due enquiry, direct such remedial measures, including the suspension of the relevant AI System, as it deems appropriate.
The section is very explicit and there is no cause for any misunderstanding. There are specific permitted uses such as the administrative functions . But even for these, prior approval is required to be taken by vendors from an appropriate authority which will be the AI committee in the Supreme Court or respective High Courts.
It is possible that multiple software can got approved from different committees. This could have been avoided by making the approval of software from the centralized Technical Committee.
Under the prohibited uses, the use of personal data for training of the algorithm is prohibited and so also is any automated judicial decision making. This is welcome and DGPSI-AI is already in sync with this thought.
It is preferable if clarification is also made available that violations if any are reported to the central committee so that the vendor or the system can be removed from the system in all other Courts where it might have been earlier in use.
Naavi
