Court’s decision
The Madras High Court has, for the first time, formally permitted the structured and limited use of an artificial intelligence–assisted system in pending commercial arbitration proceedings. The Court approved a controlled demonstration of the “Superlaw Courts” algorithm and allowed its provisional use strictly as a record-management and summarisation tool. Emphasising that judicial reasoning, legal conclusions, and adjudication would remain entirely human-driven, the Court directed that the AI system be used only to organise, retrieve, and summarise material already on record. To ensure transparency and party confidence, the Court postponed final hearings and directed counsels to evaluate the system’s effectiveness before adjudication resumes.
Facts
The proceedings arose out of multiple interconnected arbitration-related petitions and applications between a large infrastructure joint venture and a metro rail authority concerning high-value contractual disputes. These included petitions under the Arbitration and Conciliation Act, applications relating to bank guarantees, execution proceedings, and connected commercial division matters. Given the volume of pleadings, documentary evidence, arbitral records, and overlapping issues across several cases, the Court explored the feasibility of using an AI-assisted system to manage and navigate the record efficiently. During the hearing, the “Superlaw Courts” algorithm was demonstrated before the Court with participation from both sides’ counsel.
Issues
The principal issues before the Court were not substantive contractual disputes but procedural and institutional in nature. The Court considered whether artificial intelligence could be safely deployed to assist courts and lawyers in managing voluminous records, whether such deployment could compromise judicial independence or fairness, and what safeguards were necessary to prevent misuse, hallucination, or opaque reliance on AI outputs. A further issue was how to ensure transparency so that parties and the public could understand the extent to which AI assistance influenced the judicial process, if at all.
Petitioner’s arguments
The petitioning party, represented by senior counsel, participated in the demonstration and did not object in principle to the limited use of AI as a facilitative tool. The emphasis from the petitioner’s side was on ensuring that any technological assistance remained confined to the record and did not stray into legal reasoning, inference, or evaluation of evidence. The petitioner underscored the importance of accuracy in summarising pleadings, evidence, and arbitral findings, particularly in a matter involving multiple connected proceedings and substantial financial stakes. It was also stressed that any AI-generated output must be verifiable against the original record.
Respondent’s arguments
The respondent metro rail authority similarly participated in the demonstration and expressed prima facie satisfaction with the working method of the algorithm, subject to strict safeguards. The respondent highlighted the need for transparency and traceability, arguing that any interaction with the AI system should be open to scrutiny by both sides. Concerns were raised that AI should not create summaries that subtly frame issues or influence narrative balance. The respondent therefore supported the Court’s insistence that AI assistance end at the stage of preparing a factual draft, with no role in judicial conclusions or final reasoning.
Analysis of the law
The High Court approached the matter from the standpoint of procedural fairness, transparency, and constitutional responsibility. While acknowledging the growing complexity of commercial litigation and arbitration, the Court made it clear that efficiency cannot come at the cost of adjudicatory integrity. The Court recognised that there is no statutory prohibition on courts using technological tools for administrative or research assistance. However, it drew a firm distinction between permissible assistance and impermissible delegation of judicial functions. The law, as understood by the Court, mandates that decision-making authority, appreciation of evidence, and application of legal principles must remain exclusively with the judge.
Precedent analysis
While the order did not rely on prior case law governing artificial intelligence, it implicitly aligned with established principles from arbitration and procedural jurisprudence that limit external influence on adjudication. The Court’s insistence that AI remain “record-bound,” non-inferential, and non-opinionated reflects long-standing doctrines that judicial decisions must be based solely on the pleadings and evidence on record. The Court also echoed transparency norms increasingly emphasised in constitutional adjudication, ensuring that litigants can trace how material has been considered and presented. In that sense, the order may serve as a foundational precedent for future judicial engagement with AI tools.
Court’s reasoning
The Court carefully examined the working note circulated on the Superlaw Courts algorithm. It recorded that the system operates only on documents uploaded for the specific matter, does not consult external sources, and expressly avoids hallucination by stating when information is not found in the record. The AI does not draw legal inferences, assess credibility, or express opinions. Instead, it organises documents, converts scanned material into searchable text, groups related materials, and retrieves relevant excerpts in response to targeted queries. Satisfied with these safeguards, the Court permitted a pilot use, while making it clear that reliance on AI would end once a draft factual order is prepared and verified by both sides.
Conclusion
The Madras High Court did not decide the substantive arbitration disputes at this stage. Instead, it charted a cautious and transparent path for the experimental use of artificial intelligence in court proceedings. The matters were posted for final hearing on later dates, with counsels directed to work with the AI system for a week and report on its effectiveness. The Court ordered that all interactions with the algorithm be accessible through a separate link to ensure transparency. This marked a significant procedural development, positioning the case as the first instance of structured AI assistance being formally acknowledged and regulated by an Indian court.
Implications
This order has far-reaching implications for the Indian judiciary, particularly in commercial and arbitration matters involving voluminous records. It signals judicial openness to technological innovation while drawing clear red lines around adjudicatory independence. The emphasis on record-bound operation, verifiability, and transparency may serve as a template for future AI adoption across courts. At the same time, the Court’s cautious, pilot-based approach reassures litigants that AI will not replace judges or lawyers but may, under strict supervision, enhance efficiency and accuracy in managing complex litigation.
Case law references
- Present Madras High Court order on AI-assisted systems: Held that artificial intelligence may be used only as a record-management and summarisation tool, with no role in legal reasoning or adjudication; mandated transparency and party verification before final hearing.
FAQs
Q1. Has the Madras High Court allowed the use of artificial intelligence in court proceedings?
Yes. The Court has permitted limited, supervised use of AI strictly for organising and summarising records, not for judicial decision-making.
Q2. Can AI influence judicial reasoning or outcomes in this case?
No. The Court expressly held that AI cannot draw legal inferences, assess evidence, or influence adjudication, which remains entirely with the judge.
Q3. Why were the arbitration hearings postponed?
The hearings were deferred to allow counsels to evaluate the effectiveness of the AI system and ensure accuracy and transparency before final arguments.