The Supreme Court Advocates-on-Record Association (SCAORA) has opposed the disclosure requirement in the Supreme Court’s draft AI regulation that would force lawyers to declare every use of AI tools in preparing pleadings, calling it “unworkable” and unjustified given existing professional codes. The comments come in response to the Supreme Court’s Draft Regulations for Use of Artificial Intelligence (AI) in Courts, 2026, released for consultation by the AI Committee. 

Why SCAORA opposes the disclosure requirement: Regulation 43(3) and (4) require advocates to disclose “the AI-assisted character” of any document through a formal certificate whenever an AI tool is used in its preparation. SCAORA “strongly opposes” this, arguing that:

• Advocates are “officers of the court already bound by strict statutory codes of professional conduct” under the Advocates Act, 1961, and the Supreme Court Rules, 2013.
• “The ultimate filing remains the sole responsibility of that advocate, who personally vouches for the accuracy of its contents.”
• “Requiring lawyers to constantly file declarations for using everyday software is unworkable as it can prejudice the party” or “the presiding Judicial Officer.”

Instead of blanket disclosure, SCAORA proposes amending the existing SLP certificate to insert a targeted line certifying that “all legal citations, judicial precedents, and statutory authorities relied upon in this Petition have been personally verified for accuracy.” This, it says, “would reinforce that under the Advocates Act, the lawyer is already fully responsible for their pleadings.”

Hallucinations and the “Black Box” problem: SCAORA’s technical critique flags “four fundamental concerns” with AI’s suitability for judicial use. On hallucinations, it invokes the Supreme Court’s own warning in Pooja Ramesh Singh v. Jammu & Kashmir Bank Ltd., where the Court likened fabricated AI precedents to “the release of methyl isocyanate in the province of law and justice: invisible, insidious, and catastrophic by the time anyone notices” and that “it is necessary for Courts to adopt a zero-tolerance mode for producing, citing or using AI-generated precedents without verification”.

On explainability, SCAORA notes that “the explanation furnished is another output,” not “a record of the actual internal process by which the output was generated.” It adds that no AI system “is immune” from the Black Box paradox, since “the output is generated from the interaction of a vast neural network” and “there is nothing like fact A, legal principle B and conclusion C.”

SCAORA also challenges Regulation 20(1)(c), which bars AI from adjudication “without mandatory Human-in-the-Loop.” Read literally, it says, this “suggest[s] that AI may perform adjudication or sentencing so long as a human is placed in the loop,” warning that “mere presence of a human cannot be allowed to legitimise AI-based adjudication.”

Governance and data: The report criticises the “pyramid type of governance system” spanning the Apex Body, five Standing Committees, CoRE-AI, AI Committees, and an AI Content Verification Authority, warning this “creates the possibility of repetitive evaluations of the same AI system” and makes it “difficult to fix accountability.”

On data, SCAORA warns that “the data of millions of Indian citizens cannot be exposed to data centers of large language models controlled by trillion-dollar multinational corporations located outside India.” It further argues that Regulation 47’s bare cross-reference to the DPDP Act is inadequate, since the Act and Rules are “being brought into force in phases over eighteen months,” with several key obligations “scheduled to become enforceable only on 13 May 2027.”

Also read: