Ratio Chat

The citations were real. The propositions were not.

A ₹425 crore penalty, set aside. In the Court's words: “Some of the case laws relied upon … which do exist, do not lay down the ratio deduced therefrom and it appears to be a hallucination of AI.”

Vijay Ghanshyam Gadiya v. Union of India, 2026 INSC 947

Sign in to start

No password — we email you a link. The first sixty questions are free.

Ratio has two ways in: ask it here, or search the corpus yourself.

Three ways an afternoon disappears

None of them are about not knowing the law. They are about the distance between knowing a case exists and being able to put it in a filing.

You know there is a case

You can describe the proposition exactly. You cannot remember which bench said it, or when. So you search by words the judgment may never have used, and read eleven that are close.

You find it. It has moved.

The authority is sound and three later benches have distinguished it on facts closer to yours than to the original. Nothing on the page you are reading says so.

Or the tool invents the holding

Not the case — the case is real. The ratio is not. You quote it, opposing counsel reads the judgment, and the paragraph says something else.

What an answer looks like

Every authority the answer names appears beside it, as our index holds it — so the claim and the record are on the same screen.

Conversation
Is Bhajan Lal still good law on quashing an FIR?
Yes — it has not been overruled. The seven categories in State of Haryana v. Ch. Bhajan Lal, 1990 INSC 363, remain the starting point. Its recorded treatment is distinguished by 3 later benches against 51 relying on it.

What has tightened it in practice: Neeharika Infrastructure, 2021 INSC 253 — quashing “should be an exception rather than an ordinary rule”.
Sources

State of Haryana v. Ch. Bhajan Lal

1990 INSC 363 · cited by 117 benches
Status Distinguished — held not to apply by 3 later benches Ratio
“…the power of quashing should be exercised very sparingly and with circumspection and that too in the rarest of rare cases…”

Why you are not taking its word for anything

Every case in that order would have passed a citation-checker. They existed. What did not exist was the ratio deduced from them — and a list of citations will never catch that. So the record this worked from sits on the screen beside what it said about it, and you read both.

It has no memory to answer from

It is given no case law. Everything it tells you about a judgment comes from a lookup in our index during that answer, and where the corpus holds nothing on point it says so rather than assembling a plausible authority.

The record is shown, not described

The panel beside the answer is built from our index, not from the model's words. A citation we do not hold is listed as not found — the discrepancy is on screen rather than hidden.

The ratio, verbatim, with its paragraph

Quoted from the judgment, not summarised into quotation marks — which is the exact failure in 2026 INSC 947. Cited as 2021 INSC 253, never a private reporter, and linked to the Court's own PDF.

Good law, as recorded

Where later benches distinguished, doubted or overruled a case, the answer says so — and can name which benches, and when.

Start with sixty questions on us

A new account opens with 300 credits and another 300 when those run out — about sixty questions, or six hundred searches in Ratio. No card, no subscription, nothing monthly.

Sign in to start

No password — we email you a link.