Last week, the Delhi High Court passed an important judgment in Laksh Vir Singh Yadav v. Union of India. Justice Sachin Datta held that the right to privacy under Article 21 extends to the right to be forgotten. The relief granted to the petitioners asked search engines like Google and service providers like Indian Kanoon to delist names from judgments and orders and to disable name-based searches. The judgment aims to retain the public nature of the records and to discourage access to them through name-based searches.
This is a reasonable ask. The European equivalent of this is Article 17 of the General Data Protection Regulation (GDPR), under the right to erasure. The moral argument is simple and intuitive. Humans forget when machines do not. A person is entitled to some distance from their worst day. This is the right to be left alone, restated for a medium that has abolished the natural decay of memory. Nobody needs to be reminded ad nauseam of an arrest that ended in acquittal, a complaint that ended in a closure report, or a marriage that ended in court. A person mentioned incidentally in litigation should not spend the rest of their life explaining a case that was never really about them. Internet prevents this forgetting.
On that normative question, the courts’ views echo common sense. They also acknowledge the importance of open justice. Records are to remain open and searchable for public interest. For instance, an offender against women or children gets no clean slate, because the public interest in that information does not fade with time and exists to protect other people. Likewise, a person convicted of a breach of public trust or an elected representative faces heightened and continuous scrutiny under the logic of public accountability. The courts sorted cases by their public relevance, so that a criminal conviction may legitimately surface on a name search while a matrimonial dispute may not, and it allowed masking of private and sensitive matters on application. The rule against de-indexing aims to deter only easy aggregation.
However, it is hard to take a right seriously that is granted but not guaranteed. Intent alone is wishful thinking if not aided by sensible policy that is cognisant of real constraints. The process surrounding the right to be forgotten needs to be practical, and implementable. The Delhi High Court judgment does not deliver on this expectation.
Start with the definition. Who is the forgetter? The judgment names Google and Indian Kanoon, and then orders the Ministry of Electronics & Information Technology (MeitY) to communicate directions to named parties and all other search engine operators and intermediary platforms operating within India. When MeitY files the compliance affidavit in four weeks, will it contain a list of all such providers? Or will MeitY put out an instruction to the public at large? There is no stable, obvious universe of actors waiting to be notified — because the judgement does not really engage with the question of the forgetter.
Most people do not consume a case by reading judgments, or searching across legal databases. They read news reports, tweet threads by LiveLaw and Bar and Bench, social media opinions by lawyers. Google may have become a common noun, but they are not the only search provider. The High Court judgment recognises that “associate reportage” to a court order exists. What is the expectation on these entities?
The limits of de-indexing
The court should contend more deeply with Shakespeare’s question. “What’s in a name?” It assumes that there is a clear and identifiable party field in court data. Those who have wrestled with High Court and District Court orders recognise that most courts hydrate the case title with “And ORS” (And Others), making them of limited use. Orders are poorly scanned and are most often littered with misspellings and incoherence. Private search service providers clean up entities from court metadata and aggregate such information to make it searchable. Turning off party-name searches is easy, but cosmetic. Even if names are de-indexed, what prevents a regular, free-text search for a given name? It would still match on the order text and hardly solves for the problem the court tries to address.
A regulation that can be evaded by anyone with enough resources should be honest about its limitations. For example, we have a Do Not Disturb Registry. Despite being on it, most people still receive spam calls peddling credit cards, loans, health, wealth, and other shady things. It is but one example of a rule not being properly enforced. Indian platforms with offices in the country, who fear being in contempt of the court or government action, may comply. But the internet is global. Rules imposed by Indian courts cannot bind actors sitting outside India’s borders beyond a point.
Moreover, most things on the internet have already been copied into some or the other database. Apart from legal databases like Manupatra and SCC Online, there are large language models that have already ingested huge volumes of public court records. Once a name sits inside a training corpus, tokenised and stored in a vector database, de-indexing Google achieves little. Even prospectively speaking, while this judgment may have made it tougher to use code and aggregate appearances of a person’s name, it is not too expensive to hire people to do such tasks manually.
Also read: Indian courts are operating like a lottery. They need smarter case scheduling
Fixing the source
Finally, there is an odd gap between the court order and the source of the problem. For some reason, the attention is on Indian Kanoon when they are merely republishing what the courts do. The Supreme Court’s own case-search system permits party-name searches, with the practical limitation that the search be year-wise.
In a matter referenced in the judgment, the court notes that the identity of a victim of sexual offences was statutorily protected under Section 228A of the IPC (and 72 of the BNS, 2023). It then goes on to chastise Indian Kanoon, casting liability upon them to “verify” at “the point of upload” that no identifiers are present in the case orders. This censure should have been directed against the court’s own registry, which published the record, rather than placing the onus on a private data provider.
The greater irony is that firms like Indian Kanoon (and us) exist because official alternatives are poor. The eCourts portals, individual High Court and tribunal sites and various official government databases are slow and unreliable. They are riddled with idiosyncrasy: case-type abbreviations vary by registry, sessions expire, captchas interrupt access, inconsistent formats resist portable or machine-readable use.
This is not to suggest the Delhi High Court judgment isn’t sound. That consensus is societal, and the courts have rightly identified a real harm, located it under Article 21, and refused to let the absence of a statute become an excuse for inaction. But a good diagnosis doesn’t lead itself naturally to a workable cure. Without a coherent framework on data privacy and the date of expiry on public proceedings, we are left to debate the subjective. Is this a principle? Or a discretionary cleanup that is to be conducted on a case-by-case basis? After all, the court itself rejected the reliefs sought by some of the other petitioners in connected cases.
The judgment may have laid out an intent. The execution of that intent requires detailing. This requires legislation, clear rules on data fiduciaries and their responsibilities. Who is responsible for forgetting—the court, the search engines, newspapers, databases, the artificial intelligence?
Courts also need a better framework for legal information systems. De-indexing may have its place, but it is a simplistic understanding of information storage and retrieval. While good faith actors may comply, the burden of forgetting cannot be on whoever is easiest to find. Forgetting, like justice, has to begin at the source. And it can only work if someone is held responsible for remembering properly in the first place.
Siddarth Raman is founder and CTO of TheProfesseer, a litigation analytics platform that runs courtdaily.info, a website that works on making access to court records more seamless. He tweets @thriddas. Gokul Sunoj is an associate at The Professeer. He tweets @GokulSunoj. Views are personal.
(Edited by Ratan Priya)