Cloneable versus unclonable: how secure NFC authentication works

Most of the product authentication on the market today can be copied by anyone with a phone and a few minutes, because it is a static code that says the same thing every time it is read. Secure NFC authentication works differently: the tag computes a fresh, single-use signature on each tap, so a copy of a previous read is worth nothing.

That distinction is the whole subject. Everything else is implementation detail, and a few uncomfortable limits that suppliers tend not to lead with.

Why static codes fail

A printed QR code, a serial number, a barcode and a basic NFC tag carrying a web address all have the same property: what they contain is fixed, readable by anyone, and identical on every read.

So the attack is trivial. Buy one genuine item. Read the code. Print it, or write it, onto fifty thousand fakes. Every one of those fakes now passes the check, because the check was only ever asking whether the code was on the list of valid codes.

It gets worse with plain NFC. An unprotected tag can usually be duplicated onto a blank tag in seconds with a consumer phone app. The counterfeiter does not need to understand anything. They need one genuine product and a device that costs nothing.

This matters because a static code still looks like security to a customer. The pack says scan to verify, the page says genuine, and the customer has no way of knowing the same page is being served to thousands of counterfeit units.

The real question is not “is this code valid”

It is “is this the only item using it”.

Static systems cannot answer that. Every clone is indistinguishable from the original, so the system sees one product that is scanned unusually often, which is exactly what a popular genuine product looks like.

Secure authentication changes the question to one the tag itself can answer.

What an unclonable tag does differently

No mathematics, just the mechanism.

A secret is put inside the chip and never leaves it. During manufacture, each tag is given its own key, held in tamper-resistant silicon. The chip will use the key to perform a calculation. It will not read the key out, to a phone or to anything else. Attacking the chip physically to recover it is expensive, destructive and generally not worth doing for one tag.

Each tap produces a new code. The chip keeps a counter that increases on every read. When a phone taps the tag, the chip combines the counter with its secret and produces a short signature that has never appeared before and will never appear again.

The server checks it, then burns it. The signature is sent with the tag’s identifier. The platform holds the matching key, recalculates what the answer should be, and compares. It also checks the counter has moved forwards. A previously used signature is rejected, so a captured read cannot be replayed.

Copying is therefore pointless rather than difficult. A counterfeiter who reads a genuine tag a thousand times gets a thousand signatures that have all already been spent.

What it still does not prove

Here is the part that gets skipped in most sales decks, and it is the single most important thing to understand before buying anything.

An unclonable tag proves the tag is genuine. It does not prove the product is.

If the tag is a label, it can be peeled off a genuine item and applied to a fake. If it is in the packaging, the packaging can be refilled. Neither of those attacks needs to break any cryptography, because they do not attack the cryptography.

Which is why serious deployments spend as much attention on the physical bond as on the chip: tags built into the product rather than stuck on it, adhesives that destroy the antenna on removal, tags under a seal that must be broken to open the item, and tags placed where the counterfeiter would have to rebuild the product to reach them.

The second limit is human. Authentication only works if somebody actually checks. A tag nobody taps protects nothing, which is why the consumer-facing reason to tap tends to decide whether an authentication programme succeeds. Provenance, ownership registration, refill credit, warranty: something worth the customer’s five seconds.

What you can detect without cryptography

Worth knowing, because it is cheaper and often deployed first.

Even a static code produces useful signals once you log every scan. The same serial read in Leeds and in Guangzhou within the hour. A code from a batch shipped to one distributor turning up consistently in another country. A single identifier scanned four hundred times when the average is three. Scan velocity, geography and distribution pattern will surface a clone operation, just after the fakes are already circulating rather than at the first tap.

Anomaly detection tells you that you have a problem. Cryptographic tags stop an individual fake at the point of sale. Most programmes end up wanting both, and it is reasonable to start with the first while planning for the second.

Questions worth asking a supplier

• Is the key generated inside the chip, or generated elsewhere and loaded in? Where is the key database held, and who else has a copy?
• What happens if verification is offline or the phone has no signal?
• Can the tag be verified by a standard phone with no app installed?
• How is the tag physically bonded to the product, and what happens to it if somebody tries to remove it?
• What is the plan for the tags already in the field when a chip generation is eventually broken?
• Can you show me a clone being rejected, on real stock, today?

That last one is the honest test. Ask for the demonstration rather than the diagram.

Where this sits

Veribli’s roots are in industrial RFID: nearly thirty years of tracking physical items through real supply chains, including over 900 million refillable containers, and an Innovate UK competition win in 2022. Tag selection, physical bonding, item-level identity and giving the customer a reason to tap are the parts of an authentication programme that most often decide whether it works, and they are the parts we can talk about from experience.