OUTLINE

1. Introduction
2. Framework of the Draft Regulations
3. Critical analysis of the Draft Regulations 

1. Rule of Law
2. Judicial Independence and Separation of Powers
   1. Vulnerability to corporate capture of the judiciary
   2. Subversion of democratic process
3. Procedural justice
4. Accountability
   1. Scope for arbitrariness–undefined consequences and excessive discretion
   2. Unjustified presumptions
   3. Practical concerns
   4. Misplaced accountability burdens
   5. Challenges in holding experts accountable
   6. Procurement and engagement with Private Entities: Transparency Concerns
5. Access to Justice
6. Implications for the Legal Profession

IV. Concluding remarks
V. Way forward and Recommendations

 I.       Introduction

UNESCO’s 2024 global survey on judicial uses of Artificial Intelligence (AI) systems revealed that though 44% of Respondents judges use ChatGPT and others for work purposes, only 9% of surveyed judicial officers had received any AI-related training or information; and 73% of the respondents considered that there should be mandatory regulations and guidelines for the use of AI tools in the judiciary.

The Supreme Court of India (“SCI”) has been concerned about Advocates citing non-existent cases^[1] and has set aside orders on the ground that “a decision of a Court or an adjudicating authority based on material which is fake and hallucinated is no decision at all, and it amounts to subversion of the rule of law”^[2].

The Draft Regulations of Use of Artificial Intelligence in Courts, 2026, (“Draft Regulations”) published for public consultations on 5th June 2026, in this backdrop, is indeed a welcome intervention by the apex court, intending to enable responsible procurement and use and integration of AI in courts and judicial processes.

The European Union’s Artificial Intelligence Act classifies use of AI in the administration of justice, i.e., to assist a judicial authority in researching and interpreting facts and the law and in applying the law to a concrete set of facts, or to be used in a similar way in alternative dispute resolution, and democratic processes as a high risk endeavour^[3].

Yet, the SCI’s urgent striving towards adapting AI without undertaking a study of present use, implications and cost-benefits analysis is concerning. In the legal fraternity’s collective  enthusiasm about AI, we must not ignore the critical and reflective processes that illuminate the normative framework and politics underpinning such technology.

Through this response paper, we attempt to assist the legal fraternity as well as other relevant stakeholders in understanding, engaging and deciding on whether and how AI can be used in the justice delivery systems across India, particularly from a perspective informed by our engagement with law from a democratic and human rights centric approach.  We aim to accomplish the task by, firstly, analysing whether and how the general principles provided in Regulations 4 to 17, that are to govern the adoption, deployment and use of AI the judicial system, are operationalised by the Regulations 18 to 57.

Secondly, we hope to critically examine the possible implications of these regulations in light of the broader socio-political, normative and perhaps philosophical questions that the judiciary must engage with while it balances its constitutional responsibilities with its goals towards efficiency, speedy disposal and accuracy.

   II. Framework of the Draft Regulations

At the outset the Draft Regulations in Chapter II lays down the general governing principles  for the adoption, deployment and use of AI in the judiciary, which are — human primacy and judicial independence, rule of law, fairness and non-discrimination, transparency and explainability, accountability, auditability and continuous oversight, inclusivity and accessibility, purpose limitation, proportionality as some general principles that are to govern the adoption, deployment and use of AI in courts. It also envisions safeguards through data privacy and protection, data integrity and cyber security while introducing presumptions in favour of responsible AI adoption and an encouragement towards innovation over restraint.

While the hat-tip to these critical principles seems reassuring, a closer reading of the substance of these principles as articulated in the regulations is confusing if not concerning, given the role of the judiciary in the constitutional framework.

Chapter III thereafter enlists permissible (with human oversight and verification) and (absolute) prohibited uses of AI while conferring the overarching power to determine utilities to the Appropriate Authority (Reg. 18) which could be the Apex Body in the Supreme Court or the AI committees in the High Courts and tribunals.

Regulation 21 grants the  AI Committees powers to take “remedial measures as it deems appropriate” for violation of Regulation No. 20.

Chapter IV outlines the institutional mechanism for governance, oversight and policy making. It constituted an Apex Body with six committees for assistance in discharging its functions. There is also an AI Committees to oversee, regulate and facilitate the responsible adoption and governance of AI within its jurisdiction. AI Secretariats are established to assist each of these AI Committees. Each of these committees shall be for assisting the Body in its functions.

Permissible Uses (Reg. 19)

Prohibited Uses (Reg. 20)

(a) case management – listing, docket prioritization, identifying defects in new filings, etc.   (b) Transcription   (c)  Translations   (d)  Preparation of legal briefs and legal research   (e) administrative functions  such as judicial resource allocation, defect scrutiny, records management etc.   (f) Assisting litigants and other stakeholders to access court services.   (g) Enabling access to persons with disabilities and linguistic barriers.   (h) document authenticity verification and fraud detection in administrative processes   (i) anonymisation of judgments, orders and Court records for publication   (j) analytical tools for judicial administration, court performance assessment and  backlog monitoring and management; and   (k) automated generation of administrative documents – such as court notices, summons, etc.

a) Training, testing, or refining any AI  System using personal data of any person without the prior approval of the Appropriate Authority or against the applicable data protection laws.   (b) Determining judicial outcomes (including any judgment, order, or finding of fact or  law) through Algorithmic Decision-Making alone or solely on the  basis of AI-generated information, data, or analysis.   (c) Adjudication and sentencing in any matter without mandatory Human-in-the-Loop (it can be treated as advisory)   (d) Predictive adjudication, risk scoring of flight risk, recidivism, credibility of parties, evaluation of eligibility for bail.   (e) Use of undisclosed, opaque, or unexplainable AI System in any Court process that “materially affects” rights and liberties of any parties;   (f) Prediction and profiling or inferring the future conduct or behaviour of parties, accused persons, witnesses, or legal representatives in any Court process;   (g) 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) Submissions of AI-generated outputs to a Court as an independent source of evidence without full and transparent disclosure.   (i) Use of AI Systems in any manner that may compromise the confidentiality of judicial deliberations or the independence of the judicial decision-making process.

 

Chapter V outlines the precautionary measures for oversight, audits and incident management which includes transparency through disclosure of use of AI in preparation or submission of any document, pleading, or evidence, the AI-assisted character through declarations. There is also an AI content verification authority that is tasked with the responsibility of ensuring disclosure of use of AI in court submissions. Chapter VI deals with procurement of AI systems and private sector engagement. Chapter VII with data protection and cyber-security. Chapter VIII with capacity building and training of personnel in technical, legal and ethical dimensions of AI. Chapter XI deals with grievance redressal for harms caused by an AI incident, where the Court in which the relevant AI System was or is being may pass appropriate orders as it may deem fit after giving a reasonable opportunity of being heard.

   III.          Critical analysis of the Draft Regulations

While a superficial reading of the Draft Rules indicates the use of all the right language that gives it a sheen of being progressive, a closer reading of the Draft Regulations reveals that the governing principles enshrined in Regulations 4 to 17 of Chapter II are not satisfactorily operationalised in the later part of the Draft Regulations. Moreover, it is interesting to note that human dignity does not feature in the list of governing principles. We aim to evaluate the Draft Regulations against four overarching normative and practical concerns – Rule of Law, Separation of Powers, Procedural Justice, Accountability, Access to Justice and Implications for the Legal Profession.

1.   Rule of Law

Rule of law primarily entails that laws must be accessible and as far as possible, be intelligible, clear and predictable^[4]. However, the Draft Regulations do not clearly state the liability or consequences for harm caused by an AI incident or AI security and/or data breaches or even the violation of the Regulations. In fact, in Part VI of the Draft Regulations, it relegates consequences and liability to be negotiated through contracts with private entities that are expected to develop AI systems for courts. Second, the determination of consequences for violation of prohibited uses and/or liabilities and remedies for breach are left to the discretion of the appropriate authority. Such provisions create much scope for arbitrariness, and thus, vitiate the Rule of Law requirement that law be predictable and that questions of rights and liabilities to be resolved by application of law and not exercise of discretion.^[5] Further, the Court also relies on other substantive laws to take care of consequences. However, a field as rapidly evolving and dynamic as AI, makes it difficult to ascertain whether existing laws may be interpreted to cover the various issues that may arise therefrom.

2.   Judicial Independence and Separation of Powers

a.      Vulnerability to corporate capture of the judiciary

The Draft regulation envisages significant participation of private entities in the development, procurement, deployment and maintenance of AI systems for the judiciary. While chapter VII prescribes contractual safeguards concerning confidentiality, ownership and data protection, it proceeds on the assumption that private technology companies will play a central role in shaping the future digital infrastructure of the Indian Judiciary. This raises serious concerns that extend beyond data protection or procurement processes. It raises a more fundamental constitutional question — should the administration of justice/ access to justice become largely dependent on private commercial/corporates without adequate public debate or democratic participation?

 

Given the limited domestic public infrastructure for developing advanced AI systems^[6], there is a real possibility that multinational technology corporations will emerge as the primary suppliers of AI tools to Indian courts. Over time, this risks rendering the judiciary vulnerable to a gradual corporate capture of its infrastructure, where critical functions such as legal research, transcription, translation, case management and document analysis become dependent upon these technologies owned and controlled by private companies or foreign Big Tech. Although, looking at the regulation,  there is a promise that judicial decision making may formally remain with judges, the technological infrastructure supporting that decision making process could increasingly be designed, maintained and updated by commercial/foreign entities whose primary accountability is to shareholders rather than constitutional values or the public interest.

 

Therefore, greater emphasis should be given upon developing  digital infrastructure through public institutions such as the National Informatics Centre (NIC), public universities and publicly funded research institutions. This will require the Union and the State Governments to take lead in developing the necessary infrastructure. Public ownership of judicial AI systems would better preserve judicial independence, enhance transparency, and ensure that technological development remains accountable to constitutional values rather than commercial interests.

b.      Subversion of democratic process

Integrating AI systems in the judiciary will fundamentally transform the way justice is accessed, administered and delivered with far-reaching implications for the lives of people who approach the courts for justice. While the relationship between the courts and public is significantly different from the relationship between the government agencies and public, we argue that courts, as public  institutions, must apply democratic processes while exercising their law-making powers; especially when such regulations have incidental implications for access to justice and public faith.^[7] This necessarily calls for democratic participation and public reason as a part of the process so that the people as data subjects are able to represent their interests and concerns while shaping the technological transformation of critical public institutions. Public reason requires explaining to citizens the reasons for supporting their positions in terms of the political conception of justice they regard as the most reasonable^[8].

 

This is also particularly relevant as the Draft Regulations create a presumption in favour of responsible adoption of AI in courts and mandates it without justifying the benefits and costs that such a decision may entail for the relevant stakeholders and the public.

 

The Draft Regulations provide for engagement with private entities, technology experts and vendors, yet it fails to create meaningful democratic institutional space for civil society organisations, bar associations, or representatives of communities who are also significant stakeholders affected by AI enabled justice systems. Moreover, adoption of AI systems and tools  will have varying implications based on the intersections of caste, class, gender, region, language, disability, socio-economic status etc.

 

Furthermore, the Draft Regulations also envisions an extensive reliance on “experts” to guide the SCI in the process of integration of AI into judiciary and also mandates the use of AI by officers in Regulation 16 (unless they are able to provide reasons to be recorded in writing).  It is important to note that expertise does not just generate knowledge; it determines legal and policy decisions. Such decisions affect the lives and interests of the people subject to them.^[9]

Thus, the increasing dependence of the judiciary on corporate controlled digital infrastructure also raises questions regarding judicial independence in a constitutional context. While courts may continue to pronounce judgments, the AI system through which justice is administered may increasingly be designed, maintained and updated by private entities whose primary objective and obligations are commercial in nature rather than constitutional. The judiciary roots its authority from public trust and the constitution of India, not market efficiency. Any transformation of judicial institutions must therefore prioritise public ownership, democratic participation and constitutional values over commercial interests and technological solutionism.

3.   Procedural justice

Aspects of procedural justice in respect of certain fundamental rights are imbibed in Article 20 and 21. Procedural Justice aims at producing accurate and legitimate outcomes at a reasonable cost. Having both intrinsic and instrumental value in ensuring dignity of all persons involved in a dispute is protected in the process of obtaining substantive justice, it is critical to preserve procedural justice even in the times of AI. Studies from Austria, Brazil, Canada, Estonia, Singapore, the United Kingdom, and the United States have shown that Artificial Intelligence (AI) plays a transformative role in procedural jurisprudence. It was found that though AI might enhance efficiency and consistency, it also introduces risks of bias, accountability gaps, and process failures^[10].

One of the key rights identified as an aspect of procedural justice in the age of AI is the right to explainability^[11]. While the Draft Regulations recognises explainability as a governing principle it stops short of recognising it as a right of litigants and other relevant stakeholders, limiting the right to explainability “where appropriate”, once again leaving such appropriateness to be determined by an undefined actor.

 

Moreover, the Draft Regulations fail to acknowledge or account for the reported trade-off between explainability and accuracy whereby complex AI models that achieve high accuracy often sacrifice transparency, while simpler, interpretable models may underperform^[12].

This raises the question of to what extent is it justifiable to sacrifice explainability as one aspect of procedural justice in the interest of pursuing its other aspects such as efficiency and accuracy. These are questions that require deeper deliberations and thoughts within the legal community before being required to be bound by regulations.

4.   Accountability

The normative value of accountability arrangements’ are rooted in their capacity to firstly,  enable democratic constraint of arbitrary exercise of public power; secondly, to limit abuse of power and thirdly, to facilitates enhances the learning capacity and effectiveness of the public administration by through feedback loops, appropriate incentives for officials and agencies to reflect and improve upon their policies and procedures.^[13] The accountability framework proposed in the Draft Regulations does not adequately serve any of the above purposes.

a.      Scope for arbitrariness–undefined consequences and excessive discretion

In a recent decision the Hon’ble Supreme Court has held that it is a misconduct on the part of an advocate to cite non-existent judgments without verification.^[14] The absence of such clear and definitive consequences in the Draft Regulations creates uncertainty and anxiety about using AI systems.

 

The accountability framework is also weakened by the excessive grant of discretionary powers to the officers, for instance, the power to dispense with verification for reasons recorded in writing must be justified on limited grounds to be articulated in the Regulations.

The Draft Regulations are also not straightforward about use of AI by judges, lawyers, or other stakeholders outside of the list of approved vendors who are without any good reason envisioned to be private entities only. The Draft Regulations are silent on the mechanisms for monitoring, managing and defining liabilities for harms that arise from such usage.

 

b.      Unjustified presumptions

The Draft Regulation 16 creates a “presumption in favour of responsible AI adoption” and requires every Court to actively seek opportunities to use AI unless reasons are recorded for refusing it. This approach reverses the usual burden that applies to high-risk technologies and Big Tech. Instead of first establishing that an AI system is necessary, in tandem with the constitution, the Regulation assumes and establishes that AI should be adopted unless sufficient reasons exist to reject it. Such a presumption is arbitrary in absence of comprehensive empirical evidence or study showing that AI improves judicial outcomes without compromising constitutional values.

 

c.       Practical concerns

Further, Regulation 16 does not clarify whether all courts will use a common AI system or whether different courts and jurisdictions may independently procure different AI tools. This could lead to technological division and digital divide given India having a rural demography much larger than the urban counterpart. Here, similarly situated litigants face different technological problems depending on the court before which they appear. Equality before law requires not only consistency in legal principles but also reasonable consistency in the technological systems through which justice is administered. The Draft Regulations themselves recognise risks such as bias, hallucinations, data breaches and threats to procedural fairness. In this context, introducing AI  should be done after conducting mandatory safeguards such as pilot testing, independent impact assessments, public consultation and a clear demonstration of constitutional compatibility.

d.      Misplaced accountability burdens

By creating a “presumption in favour of responsible AI adoption” and requiring judicial officers to record reasons for restricting or refusing the use of AI systems (Regulation 16, Clause 2), the Draft Regulations shift the burden from the proponents of AI system to those who seek to exercise caution while administering justice. In doing so, the Regulations normalise AI as the default mode of judicial administration without any demonstration of its necessity, constitutional compatibility or empirical effectiveness.

The Draft Regulations also envisions an extensive reliance on “experts” to guide the SCI in the process of integration of AI into judiciary and also mandates the use of AI by officers in Regulation 16 (unless they are able to provide reasons to be recorded in writing).

e.      Challenges in holding experts accountable

Moreover, experience indicates that while “experts” influence practical action despite lacking democratic legitimacy or being held to public accountability standards, due to the nature of their role being inherently deontological where they typically follow some sort of instruction while carrying out duties assigned by superiors and are usually following a legally prescribed mandate; furthermore, it is unclear who qualifies as an expert and on what basis.^[15]

 

f.        Procurement and engagement with Private Entities: Transparency Concerns

As discussed in the earlier Part 2 (B) of this paper, the imagination of Private Sector Enterprises playing a critical role in shaping the development and use of AI in courts leaves the judiciary vulnerable to unintended compromises vis-a-vis private interests. Such engagements are envisioned to be contractual in nature and it is unclear how public scrutiny is ensured, especially given that the Chief Justice has the final say on the procedures. Further, in the language of Regulation No. 46 (4), it appears as though the allocation of liability between the Court and the vendor in  the event of AI-related incidents, data breaches, or harm to litigants or third  parties is subject to negotiation based on commercial terms rather than rule of law.  Whether these will be standardised for all such private entities, or how negotiations with foreign AI suppliers will materialise, are all unclear and must be seriously considered.

 

Further, the monitoring and auditing of the AI systems and tools by the AI secretariat would depend on the technical capacity and internal expertise to independently evaluate complex privately owned AI systems subject to IP protections, especially opaque or black-box models and frontier models that may not be built in India.

5.   Access to Justice

Court Managements Systems have a very significant impact on access to justice. While AI can be deployed to improve access to justice, it takes deliberate, intentional and informed actions to facilitate such improvement smoothly. And user feedback from experience is critical for such informed decisions. A reflective study on the American court administration in 1979, revealed that public views court as “a private game preserve in which judges, lawyers and frequent litigants romp playfully while those who foot much of the bill watch from over the fence.”^[16]

Integrating AI in the judiciary without considering the needs and challenges of ordinary litigants may further impede access to justice. SCI in its enthusiastic endeavour to reduce costs for the judiciary should not increase the costs for the litigants.

Further, we must be mindful that the very nature of AI is such that the technology rewards those who engage with it and marginalises those who don’t.^[17]

And our experience representing the litigants both privileged as well as marginalised – by caste, gender and/or class, region, religion, etc. – has taught us that the digital divide in the country is severe. Some of our clients are unlettered and struggle with basic use of their smart phones. The more fundamental democratic problem is how digitization severely limits their capacity to negotiate with algorithms that control access to critical resources they might need. Any integration of AI in the court system must be developed from this awareness, to ensure AI does not become another obstacle in the paths of the poorest and most vulnerable persons’ journey towards justice; especially considering how perhaps negotiating with human agents is perhaps more equalising and dignifying than negotiating with algorithms.

6.   Implications for the Legal Profession

Global experience with AI in the legal profession reports restructuring the profession^[18]. The human judgement that makes the discretionary powers of the judiciary valuable in a constitutional democracy, is developed over time with the exercise and application of mind. Studies on the impact of AI on the legal profession have indicated that AI is automating the base of the legal profession comprising the junior most professionals threatening the profession’s capability pipeline.^[19]As David Wilkins and Anthea Roberts note^[20]:

“The commodity legal work that AI targets first — document review, legal research, due diligence — is also the apprenticeship through which domain expertise is built. Without that expertise, who will exercise the judgment that sits at the intersection? Who will know when a plausible-sounding AI output is subtly, consequentially wrong?”

 

Thus, the Bar and the Bench must anticipate such consequences and ensure human skills of judgement and capacity to reflect on legal principles is preserved through the disruptions by AI systems and tools.

Further, what does AI in legal practice mean for the attorney-client relationship in the face of mandatory engagement with AI, without certainty of foreseeable consequences, is perhaps a case for separate reflection.

But, without adequate deliberations on such issues there is a risk that AI may exacerbate existing disparities in the legal fraternity.

               IV.                Concluding remarks

It is our considered view that the Draft Regulations do not sufficiently justify the need for integration based on publicly accessible evidence.  The costs seem to outweigh the benefits without a broad understanding of the purpose of mandating adoption, deployment and use of AI systems in courts rather than mere regulation of such adoption or use.

 

The language of machinery that is often used in the context of the executive arm of the State, holds less true for the judiciary or anybody carrying out adjudicatory functions. It is more appropriate to refer to the higher judiciary in particular as the constitutional conscience of the State given its structural position and role in a democracy. Therefore, it cannot behave like the executive in the guise of efficiency or expedience. As the guardian of the Constitution, the Supreme Court of India, is subject to higher standards and a paramount responsibility towards the constitutional values and rule of law. Moreover, the faith entrusted by the public in the judiciary has a deeper basis than merely accuracy or efficiency. The Judiciary’s power and authority is rooted in legitimacy that requires that justice be seen to be done.^[21] Therefore, decisions that fundamentally reshape the administration of justice of these intersections cannot be driven solely by market actors and technical expertise, without undermining the public confidence in the justice system.

 

Lastly, beyond the other concerns that AI brings about, it also raises fundamental existential anxieties about human worth in the age of AI. Laws must be mindful of such anxieties and center its objectives in human dignity. In this regard, the Hon’ble Supreme Court may find relevant the Pope’s foresight^[22] in respect of AI human worth in the age of AI when he says:

“102. The use of AI is never a purely technical matter: when it enters processes that affect people’s lives, it touches on rights, opportunities, status and freedom. Important and sensitive decisions — concerning employment, credit, access to public services or even a person’s reputation — risk being fully delegated to automated systems that do not know “compassion, mercy, forgiveness, and above all, the hope that people are able to change,” [125] and can therefore give rise to new forms of exclusion. There are clearly harmful uses, such as the manipulation of information or violations of privacy. Yet there is also a subtler danger, for when AI systems present themselves as neutral and objective, they end up reflecting and reinforcing the stereotypes or ideological bias of their designers and developers.

103. Indeed, entrusting an algorithm in practice with the power to select who is worthy or not, without anyone bearing responsibility for that judgment, is to hand over the task of redefining the boundaries of human possibilities. In this process, political responsibility is also lost, not just empathy toward those excluded, which can, after all, be simulated. The exclusion of the vulnerable becomes cloaked in a veneer of neutrality and objectivity, against which it becomes difficult to raise objections. In this way, injustice goes unnoticed, and compassion, mercy and forgiveness — understood not as mere appearances but as real political actions — gradually disappear from view.”

 

V.       Way forward and Recommendations

1. Baseline Study: There must be a rigorous study on how AI is presently being used by all relevant stakeholders — judicial officers, registry etc. to understand how trial court judicial officers are using AI in their work, to define specific utilities of AI in administrative and adjudicative exercises and gaps.

• It should examine how AI tools are already being informally used in judicial and administrative functions, identify existing practices, understand the specific needs of courts, and assess the digital and institutional capacities of different jurisdictions- rural and urban.
• Any regulatory framework should be based on evidence of actual institutional needs rather than a general assumption that AI will improve efficiency or access to justice.
• Understanding the need.

2. Pilot Study and Phased implementation: The integration must be piloted in a few courts to understand the integration in practice.

• Based on the pilot deployment, an impact assessment must be conducted to inquire into
• whether the principles enshrined in Regulations 4-18 of the Draft Rules can be realised in the practical deployment of AI in the judiciary.
• Impact on fundamental rights
• Feasibility of benefits

3. The Preliminary Context-Based Risk Analysis (PCRA)^[23] : To assess the impact on Human Rights, Democracy, and the Rule of Law

• Following pilot deployment, an independent impact assessment should be conducted before any large scale implementation. The assessment should examine whether the principles embodied in Regulations 4 to 18 including rights, dignity, fairness, transparency,

accountability, privacy, explainability, human oversight and non-discrimination, can be meaningfully realised in practice rather than merely in principle.

• It should specifically evaluate the impact of AI on fundamental rights under Articles 14, 19 and 21 of the Constitution, with particular attention to historically marginalised communities, including Dalits, Adivasis, women, persons with disabilities, linguistic minorities, religious minorities and LGBTQIA+ persons.

4. Publication of the assessment/study in public domain: The findings from the impact assessment must be published to assist the process of stakeholder consultations – which include civil society organisations and litigants.
5. Consultation with Stakeholder

• SCI must actively engage with civil society groups that have been engaging with digital rights and AI, and bring them in as key stakeholders that are able to contribute to the shaping of how AI impacts lives in courts, at par with “private entities”, who are also shaping the hard technologies. A collaborative approach may be more effective in ensuring that use of AI is in the service of a broader and thicker concept of Justice.
• The regulations must emerge from such a process rather than a top down approach.

6. Revoke Regulation 16 entirely – as such a presumption is arbitrary and unsubstantiated by evidence.
7. Regulations should be separated from guidelines for integration: The Draft Regulations seem to conflate regulation with integration. While integrations/ use of AI must be left to the choice based on needs and benefits, regulation determines how AI systems and tools are to be used. The Hon’ble SCI’s powers are limited to regulation. But it is contestable if it is appropriate for the apex court to mandate integration and use from a top-down approach.
8. Public ownership of critical use of AI : Encourage and prioritize publicly funded, publicly governed, indigenously developed over private players to ensure transparency in procurement and accountability. Treat Judicial data as constitutional public infrastructure and prohibit its commercial exploitation or unauthorized secondary use.
9. Uniform Standard for AI system : Develop a uniform national standard for AI usage in courts to prevent unequal digital/ technological divide across the country
10. Right to access AI usage information: Wherever AI systems materially judicial processes, guarantee and provide litigants the right to access to information regarding the use of AI in the process, human review and effective remedies.
11. Social Audit: Create mechanisms for periodic social audit on its impact on human rights, dignity and constitutional values, perhaps by committees headed by Retired Supreme Court Judges, and composed of members of civil society, human rights defenders.

^[1] Anahita Sachdev, ‘India’s top court angry after junior judge cites fake AI-generated orders’ BBC (3rd March 2026) https://www.bbc.com/news/articles/c178zzw780xo accessed 6 June 2026.

^[2] Pooja Ramesh Singh v. Jammu & Kashmir Bank Ltd., 2026 SCC OnLine SC 1258

^[3] See Article 6 (2) r/w Annex III of the EU-AI Act, 2024.

^[4] Tom Bingham, The Rule of Law (London: Allen Lane, 2010).

^[5] Id

^[6] Srinivas Raghavendra, ‘India’s Economic Problems aren’t About Currency or Gold: It’s About AI’ The India Forum (11 June 2026) https://www.theindiaforum.in/forum/indias-economic-problems-arent-about-currency-or-gold-its-about-ai accessed 14 July 2026; ‘AI Summit: India providing more data to OpenAI than US; need to build own models, says Amitabh Kant’ Moneycontrol (17 February 2026) https://www.moneycontrol.com/news/india/ai-summit-india-providing-more-data-to-openai-than-us-need-to-build-own-models-says-amitabh-kant-13831438.html accessed 14 July 2026.

^[7] Frances Kahn Zemans, ‘Eye of the Beholder: The Relationship Between the Public and the Courts’, (1991) 15(2) Justice System Journal 722.

^[8]  John Rawls, ‘The Idea of Public Reason Revisited’ (1997) 64(3) University of Chicago Law Review 765 http://www.jstor.org/stable/1600311 accessed 12 July 2026

^[9] On the problem of holding experts to account, see Sanne Taekema, ‘Expert accountability and the rule of law Intertwinement of normative and functional standards?’ Technocracy and the Law: Accountability, Governance and Expertise (Routledge, Taylor & Francis Group 2021).

^[10] Srinivas MK and M Suresh Benjamin, ‘The Rise of AI in Procedural Jurisprudence: Global Innovations, Legal Frameworks, and Future Implications’ (2025) 13(9) Journal of Law and Sustainable Development https://ojs.journalsdg.org/jlss/article/view/4534/2075 last accessed on 10 July 2026.

^[11] Ljubiša Metikoš and Iris van Domselaar, ‘Procedural Justice and Judicial AI: Substantiating Explainability Rights with the Values of Contestation’ (2025) 3(1) Journal of Human-Technology Relations https://doi.org/10.59490/jhtr.2025.3.8163 accessed 12 July 2026.

^[12] Id; ‘What Are the Trade-Offs Between Explainability and Accuracy in AI Models?’ Milvus AI Reference https://milvus.io/ai-quick-reference/what-are-the-tradeoffs-between-explainability-and-accuracy-in-ai-models accessed 12 July 2026.

^[13] Mark Bovens, ‘Analysing and Assessing Public Accountability. A Conceptual Framework’ (2007) 13 European Law Journal.

^[14] Supra at note 2

^[15] Supra note 8

^[16] Paul Nejelski and Russell R Wheeler, Wingspread Conference on Contemporary and Future Issues in the Field of Court Management (Institute for Court Management 1980).

^[17] AI in the Legal Profession’ Harvard Law School Center on the Legal Profession https://clp.law.harvard.edu/ai-in-the-legal-profession/ accessed 12 July 2026.

^[18] id

^[19] id

^[20] id

^[21] Philippa Webb, AI, Justice and Rule of Law [MOOC Lectures], UNESCO 2026.

^[22] Pope Leo XIV, Magnifica Humanitas: On Safeguarding the Human Person in the Time of Artificial Intelligence (Encyclical Letter, 15 May 2026) para 67 https://www.vatican.va/content/leo-xiv/en/encyclicals/documents/20260515-magnifica-humanitas.html accessed 15 July 2026.

^[23] UNESCO, Global Toolkit on AI and the Rule of Law for the Judiciary (UNESCO 2023) https://unesdoc.unesco.org/ark:/48223/pf0000387331 accessed 13 July 2026.