Editorial policy

Last updated: 10 September 2026

Vidhipandit publishes the judgments of Indian courts and a structured case record built from them. This page says where those documents come from, what we do and do not change, what we refuse to publish and why, and what a person named in a judgment can ask us to do about it. It is written to be used, not admired: if you are on this page because your name is in a judgment and you are worried about it, skip to If you are named in a judgment.

Where the judgments come from

We do not scrape court websites and we do not buy anyone's law reports. The corpus is built from two public, openly licensed datasets of Indian judgments held on Amazon S3 in the Mumbai region: indian-supreme-court-judgments and indian-high-court-judgments. Both are read anonymously; we take the PDF as published.

Attribution. Contains information from the Indian High Court / Supreme Court Judgments dataset, licensed under CC-BY-4.0. That licence obliges us to say so, and the same attribution is stored against every individual record in our database, not merely printed once on this page.

Reproducing a judgment is lawful in India in its own right. Section 52(1)(q)(iv) of the Copyright Act, 1957 puts the reproduction of any judgment or order of a court, tribunal or other judicial authority outside copyright infringement, unless that court has itself prohibited reproduction. Where a court has prohibited it, we do not publish.

What we take from a judgment, and what we will not take

We take the court's own words. We do not take another publisher's editorial apparatus. In Eastern Book Company v. D.B. Modak (Supreme Court, 12 December 2007) the Court held that raw judgment text is not copyrightable, but that a law reporter's own inputs are -- specifically its headnotes and editorial notes, its re-paragraphing, its internal paragraph numbering and cross-referencing, and its "concurring" and "partly dissenting" tags. Those belong to the reporter who made them. Our paragraph numbering, our structure and any summary we publish are ours, generated from the primary document.

We ingest only English-language judgments. The regional-language objects in the source datasets are refused at the point of ingestion and never enter the corpus at all. This is not an editorial preference: our safety checks described below are built for Latin script, and a translated judgment would pass through them unassessed. We would rather publish less than publish something we cannot check.

How a judgment becomes a page

  1. The source PDF is fetched from the dataset and its text extracted.
  2. Where extraction is unreliable -- a scan we cannot read cleanly -- the document is flagged and withheld. An unreadable document is a document we cannot assess, and an unassessable document is not published.
  3. The case is identified: court, neutral citation, date, document type. The source dataset's own filenames carry a year but never a day or month, so the full decision date is read from the judgment's own text where it states one clearly enough to trust; where it does not, we show the year alone rather than guess at a day. Every record on this site says which of the two you are looking at.
  4. The text is put through the redaction gate described in the next section.
  5. Only a document the gate has positively cleared is marked publishable. A single component of our software makes that decision and writes that flag; nothing on the website can write it, and no editor can set it by hand.

The redaction gate, and why it defaults closed

Some identities may not lawfully be published in India even when they appear in a judgment that anyone can download. Disclosing them is a criminal offence, not a civil wrong, and the offence is committed by the act of publishing -- so the check has to happen before a page exists, not after somebody complains.

The protections we gate against:

The gate has three rules, and they are worth stating plainly because they cost us coverage on purpose:

  1. Anything not positively cleared is withheld. Silence is not a pass. If our software cannot work out which Act a bare section number belongs to, and one of the statutes named anywhere in the document is a penal code, the document is withheld -- because we cannot rule out that the unattributed section is one of the protected ones.
  2. An error withholds. If the check itself fails part-way, the outcome is "withheld", never "cleared". A gate that opens when it breaks is not a gate.
  3. Every decision records the version of the rules that made it. When we change a rule, we can and do replay it across the whole corpus rather than leaving old documents judged by superseded rules.

Two things about how we read the law here, because they explain decisions that otherwise look over-cautious. First, the test is whether an identity is discernible, not whether a name is printed. In Nipun Saxena v. Union of India the Supreme Court treated identifying details -- the State plus a distinguishing achievement, a blurred face beside named relatives and a village -- as disclosure. In a small village the name of the village alone can be enough. Second, motive is irrelevant: the bar is absolute, and a sympathetic account breaches it exactly as a hostile one does.

And two things we deliberately do not gate, because over-suppression is its own harm. Sections 74 to 79 of the Bharatiya Nyaya Sanhita, and sections 354, 354A to 354D and 509 of the Indian Penal Code, carry no statutory bar on identifying the person concerned. Withholding every judgment under those provisions would suppress lawful reporting of cases the public is entitled to read. Where there is a privacy concern in such a case it is handled editorially, on the facts, through the review route below -- not by a blanket machine rule.

A withheld document is not deleted and not denied. It sits in the corpus, unpublished, with the reason recorded, and it is re-examined whenever the rules change.

What we add to the court's text

Alongside the judgment we publish a structured case record -- case number, bench, hearing timeline, linked orders -- and, over time, summaries and explanatory articles. Three commitments govern all of it:

If you are named in a judgment

You did not choose to be published. You may have been a party, a witness, a complainant, an accused person later acquitted, or someone mentioned in passing. You can ask us to look at it, and you do not need a court order to ask.

Write to the Grievance Officer -- the contact details are on our Grievance page -- and tell us the case and what is worrying you. What we can do, in ascending order:

Every such decision is recorded in a register with the reason and, where one exists, the reference of the court order behind it -- so that it can be reviewed, explained, and if necessary reversed. We do not quietly delete things.

We will act at once, without argument, where the disclosure engages any of the criminal protections listed above. Outside those, we weigh your position against the public interest in an open record of what courts decide, and we will tell you which way we went and why. If we decline, we say so in terms you can take to a court -- we do not simply stop replying.

What we cannot do is decide your case for you. The right to be forgotten in India is genuinely unsettled: the Madras High Court refused de-indexing in Karthick Theodore in 2021 while its Madurai Bench granted it in a separate matter, and the Delhi and Kerala High Courts have gone in opposite directions. Where a court has ordered de-indexing, we comply. Where none has, we exercise judgment and tell you what it was.

Who is responsible for this

Editorial responsibility for everything on this site rests with [PENDING-OPERATOR-FACT: name of the person holding editorial responsibility], and the publisher is Anurag Choubey, a sole proprietor trading as The Knowledge Tree (GSTIN 10AHDPC5941L2ZN), whose principal place of business is B-5, A. G. Colony, Patna, Bihar 800025. See About. Complaints go to the Grievance Officer named on the Grievance page and are answered on the timetable stated there.

This policy is dated. When it changes materially, the date changes and the change is described here rather than made silently.

Reviewed by counsel: pending. This text was drafted in-house on 10 September 2026 and has not yet been reviewed by an advocate. It states what we actually do and we stand behind it; it has not had a professional legal review.