AI in Indian Courts: Can It Clear the 5 Crore Pending Case Backlog?

Somewhere in your extended family, there is probably a court case that nobody talks about anymore because it has been running for so long. A land dispute that started when your grandfather was alive. A cheque bounce matter that was filed before smartphones became common. A family partition case where half the original parties have passed away and their children now attend the hearings instead. Everyone knows the dates keep getting pushed, and nobody quite remembers when it started feeling normal.

That feeling of normal is worth pausing on, because it reflects a number that is genuinely hard to picture. In July 2026, the Law Minister told Parliament that roughly 5.64 crore cases were waiting to be decided in Indian courts, a figure drawn from the National Judicial Data Grid, the government's own live count. And now, a new question is being asked with real seriousness: can artificial intelligence actually help clear a pile this large, or does it just add a new kind of risk to a system that is already struggling?

How Big the Backlog Really Is

Before getting into AI at all, it helps to understand the shape of the problem. The exact number moves daily and different reports cite different dates, which is why you will see figures ranging from roughly 5.1 crore to 5.6 crore depending on the source. What stays consistent across all of them is the scale and the direction. One analysis found India's pendency was growing by roughly 60,000 cases every month, with the Supreme Court's own pending count rising by over 10,000 in a single year to about 92,828 by the end of January 2026.

The burden also isn't spread evenly. According to NJDG data summarised in a policy analysis, nearly nine out of every ten pending cases sit in district courts, with criminal matters making up the majority and a significant share unresolved for more than three years. The same analysis of the backlog reported that roughly 46 lakh district court cases have been pending for over a decade. The Supreme Court alone has been reported to hold 26 matters that are more than thirty years old, while the High Courts together hold over 80,000 such cases.

What AI Is Already Doing Inside Indian Courts

AI in the Indian judiciary is not a future experiment. A few specific tools are already in use or in pilot. SUVAS, the Supreme Court's translation software, converts judgments from English into Indian languages, a quietly significant fix in a country where most citizens have never been able to read what the highest court decided about their own case in a language they actually speak. One 2026 industry listing reports it has translated over 36,000 Supreme Court judgments into the 22 scheduled languages, though that figure comes from a secondary source rather than an official court statement, so treat it as indicative rather than exact.

SUPACE, the research assistance portal built to help judges find relevant precedents, is the other well-known tool, and its story is more sobering than the headlines suggest. A 2026 research foundation analysis found that even five years after its launch, SUPACE remains in an experimental and pilot phase, with deployment limited mainly to select criminal matters before judges of the Bombay and Delhi High Courts, and with hardware, not software capability, named as a principal constraint on wider rollout. The Supreme Court has also worked with IIT Madras on an AI tool that flags defects in electronic filings, which according to reporting is accessible to only about 200 Advocates-on-Record so far.

The money behind all this is modest relative to the problem. Phase III of the e-Courts project carries an outlay of ₹7,210 crore, of which ₹53.57 crore is earmarked specifically for AI and blockchain integration. That is less than one percent of the total, which tells you where AI actually sits on the priority list right now.

A Realistic Scenario: Where AI Genuinely Helps an Ordinary Litigant

Consider an illustrative case, built from the kinds of situations that are extremely common. A small shop owner in a mid-sized town has a cheque bounce case pending in the local magistrate court. His lawyer sends a brief update after each hearing, mostly in English legal shorthand he only half understands. The case file runs to dozens of pages, and the order from the last hearing mentions several sections of law he has never heard of.

This is exactly where AI can be genuinely useful, not in deciding his case but in helping him understand it. Translating an English order into his own language, explaining in plain terms what a particular section means, summarising what happened at the last hearing, or helping him prepare a clear list of questions to ask his lawyer. None of that clears a single case from the backlog, but it reduces a quieter problem that sits alongside it: ordinary people spending years inside a legal process they do not actually understand.

The same logic applies at the court level. The pending-case problem is largely a capacity problem, and AI does not create more judges. What it can plausibly do is take over repetitive, time-consuming tasks, translation, transcription, searching for precedents, spotting defects in filings before they cause an adjournment, so that the limited hours judges and court staff have go toward actual decision-making.

The Risk Nobody Can Ignore: When AI Invents the Law

Here is where the optimistic story gets complicated, and where India's courts have already had some genuinely uncomfortable experiences. AI systems can produce what are called hallucinations, confident, well-formatted answers that are simply false, including case names, citation numbers and legal reasoning that do not exist. This has already happened inside Indian courts, not just in foreign headlines.

In one widely reported matter, a trial court order in a property dispute was found to rely on non-existent, AI-generated case citations. According to reporting on the case, the losing party challenged it in the Andhra Pradesh High Court in January 2026, and while the High Court reasoned that real property law supported the outcome anyway, the Supreme Court took a far stricter view. In February 2026 it stayed the order, and per a compilation of Indian AI hallucination cases, declared that a decision built on fake AI-generated judgments is not an error but "misconduct" carrying legal consequences.

The same compilation reports that on 2 July 2026 the Supreme Court issued a ruling declaring zero tolerance for AI-generated fake precedents, describing the situation as yet again a case of a tribunal relying on hallucinated material. The problem was not confined to one level. Fabricated citations have surfaced across tax, insolvency, trial courts and High Courts, which suggests the risk travels easily through a system where busy officials are tempted to trust something that looks authoritative.

How India's Courts Are Responding With Rules

Indian courts have started building guardrails, and the pace has picked up. In July 2025, the Kerala High Court issued what is widely regarded as the first binding AI policy for a district judiciary in India, keeping AI as an assistive tool, requiring human verification of AI-generated citations and translations, and warning against feeding case data into public cloud AI tools. The Gujarat High Court followed in April 2026 with its own policy placing responsibility for AI-assisted outputs on judges and court officers.

At the top, the Supreme Court released draft Regulations for Use of Artificial Intelligence in Courts, 2026 in June, inviting public comment. Per reporting, the draft would bar AI from deciding cases and require lawyers to disclose when they use it. One caveat worth knowing: those High Court policies apply to judges and court staff, not to practising advocates, which is part of why the Supreme Court's wider approach matters.

Why This Actually Matters for Ordinary People

It is tempting to treat all this as an abstract debate for lawyers and policy people, but the stakes are very personal. Delay in the courts is not neutral. Each year a dispute drags on, a property stays locked, a business stays stuck, a person on trial stays in limbo, and a family keeps spending money on lawyers. This is the same grinding delay that turns an unwritten will into a twenty-year property fight, as the piece on why most Indians die without a will showed.

At the same time, a justice system that adopts AI carelessly could make things worse in a different way, by producing wrong outcomes that look polished and are harder to spot. Speed that arrives with unreliability is not an improvement. The honest challenge for India is to get the benefits of automation in the boring, administrative parts of the process without handing over the part of justice that depends on human judgment and accountability.

My Honest Take: AI Can Help the Queue, But It Cannot Replace the Judges

I think the most realistic way to think about this is that AI is useful for the unglamorous parts of the legal system and dangerous when it is asked to do the glamorous ones. Translation, transcription, search, scheduling and defect checks are exactly the kind of repetitive work that eats up time and where an occasional error can be caught and corrected. Weighing evidence, interpreting law and deciding who is right are different. A wrong answer there is not just a typo, it changes someone's life, and a confident-sounding machine that cannot be cross-examined is a poor substitute for a person who can be held accountable.

What also needs saying plainly is that technology alone will not clear a backlog of this size. Reports consistently point to structural causes, too few judges relative to cases, patchy digital infrastructure in district and taluka courts where about four crore cases are pending and even basic facilities like online hearings are often missing, and a steady flow of new filings that outpaces disposal. AI can make the existing system somewhat more efficient. It cannot, by itself, fix a shortage of courtrooms and judges.

Practical Takeaways If You Have a Case Pending

Use AI to understand your case, not to build it. Asking a chatbot to explain what a section means or to translate an order is a reasonable use. Asking it to find case law you then cite is risky, because it may invent precedents that sound completely real. Always verify any case name or citation against an official source before relying on it.

Be careful about what you paste in. Courts themselves warn against feeding case data into public AI tools because of confidentiality risks. The same caution applies to you. Names, financial details and sensitive family information are better kept out of a public chatbot.

Track your case through the official portals. The eCourts services and the NJDG let you check status and next hearing dates for many cases, which cuts down the dependence on second-hand updates. Knowing exactly where your case stands is a real advantage that costs nothing.

Keep your lawyer in the loop on anything AI tells you. If a tool suggests an argument or a precedent, take it to your advocate rather than acting on it yourself. For the decisions that actually affect your case, a qualified lawyer remains the right person to rely on.

Where This Leaves Things

The honest answer to whether AI can clear India's court backlog is: not on its own, and not without real risks. It can translate more judgments, speed up research, catch filing errors and take routine work off overloaded staff, all of which helps at the margins. But the same technology has already produced fake law inside actual Indian courtrooms, and the Supreme Court's response, treating that as misconduct, shows how seriously the system now takes the danger. The most likely good outcome is a slow, careful one, where AI quietly handles the paperwork and humans keep the decisions.

Frequently Asked Questions

Q1. How many cases are actually pending in Indian courts right now?

The count changes daily and different reports cite different dates, but it sits above 5 crore across the Supreme Court, High Courts and district courts. In July 2026 the Law Minister told Parliament the figure was about 5.64 crore, based on the National Judicial Data Grid. Nearly nine out of ten pending cases are in district courts, and criminal matters make up the majority.

Q2. Which AI tools are Indian courts actually using?

The Supreme Court uses SUVAS to translate judgments into Indian languages and has piloted SUPACE for legal research, along with transcription and legal analysis tools. SUPACE remains in a pilot phase, limited to select criminal matters before certain High Court benches as of early 2026. An AI tool for catching defects in e-filings, developed with IIT Madras, is also in limited use.

Q3. Can AI really clear the backlog of cases on its own?

No. The backlog is mainly a capacity problem, with too few judges and patchy digital infrastructure in lower courts, and AI does not add judges or courtrooms. What it can do is take over repetitive tasks such as translation, transcription and research so that the limited time of judges and staff goes further. Only a small share of the e-Courts budget is currently earmarked for AI specifically.

Q4. What happens when AI makes up fake case citations in court?

Indian courts have started treating it very seriously. In February 2026 the Supreme Court took up a trial court order in a property dispute that relied on non-existent AI-generated citations and described such reliance as misconduct rather than a simple mistake. A further ruling on 2 July 2026 reportedly declared zero tolerance for AI-generated fake precedents.

Q5. Can an ordinary person use ChatGPT or similar tools for their own court case?

It is reasonable for understanding, such as translating an order or getting a plain-language explanation of a legal term, but it is risky for finding legal precedents because AI can invent cases that sound real. Any citation it gives should be checked against an official source, and sensitive case details are best not pasted into public tools. For decisions that affect the actual case, a qualified lawyer should be consulted.

This tension between AI's promise and its reliability shows up across other high-stakes areas too. AI Detectors Are Wrong More Often Than You Think looks at what happens when an automated tool is trusted too far in an accusation, and Deepfake Videos of Indian Politicians and Celebrities covers another area where trust in what AI produces is being tested.