More than one in three milk samples fails India's safety standards, and the feed that contaminates it is unregulated

India

What is happening

In 2025-26, state food safety authorities analysed 39,351 samples of milk and milk products. 14,668 of them — 37.3%, more than one in three — did not conform to the standards set under the Food Safety and Standards Act. The year before it was 12,780 out of 33,405. These are the government's own provisional figures, tabled in the Lok Sabha in July 2026.

Across every food category that year, 40,023 of 2,23,808 samples failed. Milk and dairy produced over a third of all the failures in the country from under a fifth of the samples tested. No other food category comes close.

Why this one is about children

Milk is not one food among many for a child. It is often the single largest thing a growing child consumes by volume, and every contaminant limit in the world is set per kilogram of body weight. The same glass that is a small exposure for an adult is a large one for a five-year-old. A level that is tolerable in an adult diet is not automatically tolerable in a child's — and milk is precisely where that difference lands.

What is actually in it

The word "adulteration" makes people picture detergent and urea. FSSAI's own National Milk Safety and Quality Survey, which tested 6,432 samples, found close to the opposite. Deliberate adulteration was rare — 12 samples. The real safety failures were contaminants that no shopkeeper adds:

• Aflatoxin M1 above the permitted limit in 368 samples (5.7%). It appears in milk when a cow eats mouldy feed. Aflatoxin B1, which it derives from, is a Group 1 human carcinogen. It was found more often in processed milk than in raw milk.
• Antibiotic residues above permitted limits in 77 samples (1.2%) — a direct contributor to the antimicrobial resistance that then turns up in paediatric wards.
• 41% of samples failed quality standards on fat or SNF while remaining safe to drink, which usually means dilution with water.

That distinction matters, because it changes the fix. You cannot inspect aflatoxin out of milk at the shop counter. It arrives in the cow.

The hole that just opened

In 2019 FSSAI said plainly that aflatoxin M1 enters milk through "feed and fodder, which are currently not regulated in the country". In 2026 that got worse rather than better.

In Godrej Agrovet Ltd v FSSAI (W.P.(C) 1079/2025), the Delhi High Court quashed Note (c) to Regulation 2.5.2 and the 2019-2021 directions that had required BIS certification for commercial cattle feed. The reasoning was jurisdictional and, on its own terms, sound: the FSS Act 2006 governs food for human consumption, animal feed is excluded from the definition of "food", so the Food Authority never held that power. The Court also noted that making a BIS standard mandatory requires a Central Government notification under the BIS Act 2016 — and for cattle feed, none was ever issued.

So the position today is this. The contaminant limit in milk is enforceable. The feed that puts the contaminant there is not. IS 2052:2023 caps aflatoxin B1 in compounded cattle feed at 20 ppb, and it is voluntary. The regulator that identified the cause has been told it may not address it, and no other authority has been told to.

The limit itself

India permits aflatoxin M1 in liquid milk at 0.5 µg/kg. That is the Codex level, so it is defensible on its face. The European Union permits 0.05 µg/kg — ten times stricter — and 0.025 µg/kg in infant formula.

India applies that same 0.025 µg/kg to infant formula. So the gap is only in ordinary liquid milk: the milk a two-year-old drinks from a glass is held to a limit twenty times looser than the tin the same child drank from a year earlier. If the stricter number is right for formula because infants consume it, the reasoning does not stop at the edge of the tin.

What enforcement actually produces

Of the 14,668 dairy failures in 2025-26: 10,714 civil cases and 1,054 criminal cases were launched. 9,783 civil cases ended in a penalty. 204 criminal cases ended in a conviction. Civil penalties totalled ₹34.37 crore — an average of roughly ₹35,000 per case. Criminal fines totalled ₹34.51 lakh across the entire country.

Across all foods over five years, 1,85,780 samples were found non-conforming and 6,203 convictions followed: 3.3%. For a business selling contaminated milk, the expected cost of being caught is a small and occasional fee. That is not a deterrent, it is an operating expense.

What is being asked

1. Publish the breakup. The government reports "non-conforming" as a single number that mixes adulterated, substandard, contaminated and misbranded. Parliament asked for the split and for food-wise rates, and did not receive them. Until that data exists publicly, nobody can tell whether milk is being watered down or contaminated — and those need entirely different responses.
2. Close the feed gap the Delhi High Court identified. Either notify IS 2052 as a mandatory standard under the BIS Act 2016 for feed sold for dairy animals, or amend the FSS Act to bring feed for milch animals into scope. The judgment named the mechanism. It needs the Central Government to use it.
3. Move the liquid-milk aflatoxin limit toward the infant-formula logic, on a published, staged timetable that dairies can plan against.
4. Mandatory aflatoxin M1 and antibiotic-residue testing at milk chilling and collection centres, with results published per centre. Aggregation points are where a single test covers thousands of litres, which is the only place this is affordable at scale.
5. Report child-nutrition milk separately. Where milk is supplied under mid-day meal or ICDS-type programmes, results for those supply chains should be published as their own series rather than folded into a state total.

What this post is not

This is not a claim that Indian milk is poison, and it is not an attack on dairy farmers or cooperatives. FSSAI's survey found roughly 93% of milk safe on contaminant parameters, and that deserves saying plainly. Nor is it a claim that nobody is working: inspections rose to 5,20,566 in 2025-26 from 4,01,391 the year before.

The objection is narrower. A failure rate above one in three has held steady for years. The largest identified cause sits in a part of the supply chain that a court has now confirmed nobody is required to regulate. And the penalty for being caught is smaller than the saving from not complying. Those three facts together describe a policy gap, not a law-and-order problem — and policy gaps are closed by decisions, which is why this is addressed nationally rather than to any one inspector.

How you can strengthen this

Add a response below if you have something concrete:

• A lab report on milk you had tested — the vendor type or brand, and which parameter failed.
• An RTI reply from your state food safety commissioner on dairy sample results, prosecutions launched, or testing-lab capacity in your district.
• If you work in a dairy, a cooperative or a testing lab: what is actually tested at collection, and how often.
• If you are a paediatrician: what you see, in general terms, with no patient details.

If the problem where you live is a specific dairy, vendor or collection centre, raise that as its own post against your own district or ward so the officials with jurisdiction there are the ones tagged. This post is the national one: the standard, the statute, and the notification that was never issued.