Sadly, this is unlikely to be deployed commercially. But it's cool to come up with new ideas in the space!

A couple key points:

* Readers work at a 4cm distance

* Rather than a key database, the researchers propose an AI to memorize patterns

These are no-go for almost all product ID scenarios I can imagine. Not least because the AI will have a harder and harder time validating patterns the more patterns are thrown at it. As described, a registry would be better, much better. But still bad.

David Chaum has done some work using glitter patterns on banknote bills, which I think is super cool, essentially: random reflective pattern from glitter -> some sort of FFT / visual identification signature -> stored in a registry.

Even that is going to be somewhat tough because you'll need the registry. Or alternately a chip that can store a signature and a reference to the pattern, in which case you'll just need a reader that knows the public key of the signer.

I developed some prototypes in the authentication space about ten years ago. Our approach was to detect fluorescence; first in IR, then shifted to Blue-to-Yellow shift that could be triggered by a mobile phone screen and simultaneously “read” by the front facing camera.

The fluorescing dyes were trademarkable and very hard to develop, which we compared to holograms as markers of authenticity.

As you note, inks are a big deal, and control of inks and the ink supply chain can be a big deal.

I put a bunch of money into Noteworthy a few years ago, which successfully embedded ARM chips into banknotes, the proposal being that you could at-note security with challenge/response over NFC. Noteworthy picked Bitcoin notes as a first product/demo tech. They work, it's very cool tech, and, shockingly, the chip embedding part of the tech stack isn't even the most expensive part of making a banknote.

I also invested in a company looking at drug supply chains in low-data areas, which just used SMS-based single authentication checks (great idea -- didn't pan out), a company embedding chips in sneakers (great idea -- this one works, but it's not clear if it matters for mid-tier large issuance sneakers), ...

I spent the most time with Noteworthy, and learned the banknote folks think of authentication in three tiers: individual (e.g. can you or a checkout person at a gas-station tell a note is fake), retail (e.g. can a bank teller tell a note is fake), and redacted, (e.g. can Treasury tell a note is fake). This multi-tier approach is practical and sensible to me; in brief, the farther up the chain, the more power you can bring to bear on determining a fake, and the longer you can take.

I think the same dynamics are there in a lot of consumer goods -- and the upshot is that, at least in the states with consumer goods, people mostly want to / need to trust their supply chains; it's very rare anyone is going to even take a picture of something with their phone to check its authenticity.

The MIT link doesn't explain how exactly they've done it, so that's only speculation.

The input from a scan can be quite a lot of data, a high resolution photograph, a point cloud, a set of RF timings/phases or whatever.

The key for future matching must be something small that can be distributed or sent ahead of the delivery, so a hash seems the way to go.

But standard hash algorithms are designed for non-collision and sensitivity - any changes of input lead to wildly different outputs.

But the physical product has to endure;

  - handling by humans or machines
  
  - mechanical vibration and stress
  
  - temperature variations
  
  - gravity, natural creep, thermoplasticity
  
  - bleaching due to light
 
  - variations in the reader, angle, lighting, nearby metal objects 

I don't believe their method has fine control of placement of the 'readable' parts under the glue; this cuts out methods where you're writing something in data as part of authentication. But, it also makes production MUCH cheaper - you can later just read what you "wrote" under a tag. I would guess in the price points they imagine, this method is "random + read later".

For your method - if you can write data, why not just write a digital signature of the item's serial number?

We use the pattern to mask the tag's private key. The tag authenticates itself with that private key (which is installed on the tag by the manufacturer and can be signed with a CA cert to prove it comes from the manufacturer). The tag can't decrypt the key without the correct pattern, so simply peeling it off and sticking it onto a counterfeit won't work. There is no need for a database. The AI runs on the tag and makes sure tiny changes to the pattern won't make the private key unrecoverable by outputting a "low-res" version of the pattern that is immune to such tiny changes.

That's how I understood it.

This system would probably have to be far better than 99% to be helpful. If customers test (say) a hundred thousand tags per day, we might suppose that 99% accuracy means a thousand false alarms per day. If users get suspicious and return their genuine devices upon seeing a false alarm, that's a huge cost, even without any fakes at all. Reporting "99% accuracy" allows a worrisome amount of luck into a system marketed as "tamper-proof."

Let's compare to biometric verification. Apple's TouchID claims about 1 false accept among 50,000 fake fingers (FAR=0.00002) with false rejection rates (FRR) hovering around 2%-10% or so. Even though this averages to worse than 99% "overall accuracy," it makes sense for a fingerprint matcher because the genuine user gets to retry their finger a few seconds later, so high FRR isn't a problem in practice. On the other hand, it may be better for these tags to trade off a higher FAR for a lower FRR.

I'm also worried by the article's claim that terahertz waves "travel much faster than radio waves." The speed of light doesn't depend on its frequency.

When failures approach the Per Cent-age rate, that's equivalent to ~10k DPPM, and that's daily stand-up territory.

This article does talk about this specifically:

> “One drawback is that we had a limited data sample for this demonstration, but we could improve the neural network in the future if a large number of these tags were deployed in a supply chain, giving us a lot more data samples,” Lee says.

I.e. they're well aware that 99% would need to be improved, but it sounds like they've managed to hit that level with a handful of individually created samples and a tiny project to train a model. This is just a prototype really, and 99% accuracy when testing a prototype doesn't seem unreasonable, it's not like they're planning to ship this tomorrow. It's a demonstration that recognizing these patterns automatically is actually possible in the first place, and in a serious roll-out scenario it would _probably_ be possible to train a model to do very significantly better.

it might not be like biometrics, because it could have an identifier repeated 100 times, and then the glue/tampering part could be a completely different sort of algorithm that could be optimized to physical removal, not just degradation.

Instant defeat of this device: Manufacture a tag that returns the signature of a known legitimate tag.

In order to be useful it would need to be challenge and response. Maybe they do this, but this MIT release doesn't actually give any link to a paper which is super annoying.

> Their antitampering tag contains a series of miniscule slots that enable terahertz waves to pass through the tag and strike microscopic metal particles that have been mixed into the glue.

Another issue: You would figure out how to destroy the connection between the material and the glue whilst leaving the surface of the chip intact. You would then simply re-attach it using another glue without particles in it.

> But when it came time to test the antitampering tag, Lee ran into a problem: It was very difficult and time-consuming to take precise enough measurements to determine whether two glue patterns are a match.

> He reached out to a friend in the MIT Computer Science and Artificial Intelligence Laboratory (CSAIL) and together they tackled the problem using AI. They trained a machine-learning model that could compare glue patterns and calculate their similarity with more than 99 percent accuracy.

As others have pointed out, this is a bad idea. You'd want some form of a probabilistic hash and for that to be stored in a database.

That was my first thought. If that's on a banknote, you'll have people cutting the paper around the tag and gluing it to the counterfeit, including the old paper it sticks to. This means you already need to look at the "watermark" again, you can't just read the tag.

Add some micromechanical sanding/grinding, chemical etching or laser ablation, and not only can't you read the tag with an antenna, but you need a microscope to look at - and a spectrometer, to look at the glue.

> As others have pointed out, this is a bad idea. You'd want some form of a probabilistic hash and for that to be stored in a database.

I think the reason they went with ML to do pattern recognition is because the THz response of the tag is a finicky analog signal, which might change in fundamental ways if you do things like reposition the antenna, or change what's in front/behind the tag. The problem is probably pretty similar to identifying whether two images are similar in the special case where you allow pretty significant distortions to one of the images. Hashing usually doesn't work to well for cases like that, but maybe a wavelet transform approach would work.

Exactly.

> I think the reason they went with ML to do pattern recognition is because the THz response of the tag is a finicky analog signal, which might change in fundamental ways if you do things like reposition the antenna, or change what's in front/behind the tag.

I don't doubt one second that it's difficult, but it's where the value really is. For security, you need a private key that is stored in an unreadable format that is destroyed by trying, followed by a challenge-response system to query it.

Unfortunately, without the whole picture coming together, the method cannot be adopted. With security it's kind of all or nothing. Given the value they propose it has as a solution, bypassing these checks almost definitely has value.

Strange language, this journalist uses. "Terahertz waves" are radio waves, although they're more commonly called far-infrared.

It's not clear to me what the tag itself does, apart from the glue. Apparently it serves only to hold the glue in place, and so is 100% redundant. Or perhaps the talk about "slots" explains the purpose of the tag: it sounds like a grating, but the article doesn't explain why you need a grating.

Perhaps a better solution is to create a small chip powered by electric induction. The chip would have an embedded private key and solve challenge-response queries issued by the scanning device.

I'm not sure how that compares in cost though.

Edit: it looks like these already exist and cost less than 10 cents a piece. They are called NFC tags.

>What prevents somebody from scanning it and reconstructing the position of the metal pieces?

You're talking about very, very small pieces of metal whose position/orientation is not deterministic when laying the glue and that information is combined with the tag itself to present some kind of challenge response.

If you use that, then the only way to move the NFC tag to another item would be to cut it out of the original item (including the original adhesive). But this attack also works against the technique in the article.

Regarding the orientation, I understand that it is nondeterministic in the original, but what prevents an attacker from copying it deterministically? Is it just that technology is not good enough to manipulate such small pieces of metal? How long will this limitation persist?

Perhaps in future someone could develop an advanced combination of 3d printer and pick&place machines that could do it, but such future potential doesn't disqualify this tech from currently detecting counterfeiting of fancy shoes or something.

Coat a piece of glass with a thin layer of metal. Put a photoresist on top. Project the desired pattern onto it with UV light. Wash the unhardened photoresist away and etch the unnecessary metal.

Now you've got metal in exact the spots you'd like, of exactly the thickness you'd like. You can get the accuracy down to a few hundred micrometers for cheap today.

In the case of 3D arrangements, I think some substrate materials (and also some properties of the particles) would be very difficult to get using photolithography (or some kind of micro 3D printing).

Imagine e.g. the "multiple layers of cardboard cutouts" scenery in theater vs it actually being 3D.

Assuming that doing the process you describe is sufficient, what's the ballpark of what "for cheap" means for you if you needed to print 1000 different fake tags, assuming many layers of "the desired pattern" to print the metal flakes?

If you need perfectly sharp edges and high precision, sure. But I'm sure in this case that'd be unnecessary.

> Assuming that doing the process you describe is sufficient, what's the ballpark of what "for cheap" means for you if you needed to print 1000 different fake tags, assuming many layers of "the desired pattern" to print the metal flakes?

I described in another comment an additional way to quantize the layers to reduce the repetition steps, which would reduce costs further.

Regarding costs, you could fake a THzID chip for about 500€ per fake. Not cheap enough to do it for household items, but if you're faking designer bags, clothing, sneakers, or electronics, it'd be absolutely worth it.

NFC tag usually consists of two parts very tiny IC (small piece of silicon the size of sand grain) and antenna (a piece of metal foil in a fancy shape). You could make an NFC tag where attempting to remove it rips antenna, but that wouldn't destroy the IC. It's probably a matter of product price and quantities whether, counterfeiting it by reattaching the NFC chip to a new antenna is economically viable. As the process is not only possible it's performed at the NFC tag factory at very large quantities at very low cost. It might also be possible to repair parts of broken antenna assuming area around IC is undamaged.

So overall you get simplicity and cost of regular tamper resistance stickers, with better resistance against solvent and blade attacks, and security properties closer to what you get from secure NFC chip (except you can't perform more complicated cryptographic operations like signing arbitrary data).

> Is it just that technology is not good enough to manipulate such small pieces of metal? How long will this limitation persist?

I would expect that at any point in future, whatever the best controlled manufacturing technique invented are, it will be possible to create uncontrolled pattern at finer scale, or at least much cheaper. Unless we reach the point where maintaining stable state without deteriorating becomes a problem, or the quantity of data for storing and processing becomes impractical.

I hope to not be alive when technology progresses to that point!

Nike can afford a $1 crypto tag.

to quote this thread, which someone seems determined to bury for some reason

https://news.ycombinator.com/item?id=39424773

"The issue not mentioned is one I've encountered in other similar projects which is false negatives cause by "creep" in plastics and general thermoplastic movement. We had a near-UV scanner that could map the surface of a test object (sensitive circuit boards) down to a few wavelengths, but it was too damn accurate. Go back and scan the same object a week later and stuff (like solder joints) has moved naturally. That doesn't work when things are in transit for weeks with vibrations and temperature variations"

People may be downvoting the top message because it contributes nothing. It doesn't require some nebulous malicious actor somehow trying to keep you down.

So you bought a genuine item to get its tag, don't you now have a problem of what to do with your genuine item? Sell it with fake ID?

* The attacker might have physical access to the genuine item, but not have ownership of it. A counterfeit seller might peel of stickers in a legitimate store, and use them in the fake store. Walking out with a bunch of stickers in a Farraday cage is a lot easier than walking off with the actual merchandise. The original retailer is then stuck with an unauthenticatable genuine item.

* The attacker might transplant a sticker from a defective genuine item (e.g. in a repair shop) to a counterfeit item. The genuine item was already worthless at this point.

* The attacker might transplant a sticker from an authorized object (e.g. locked suitcase that has passed inspection) to an unauthorized object (e.g. locked suitcase with contraband). The original object can then be discarded.

This tech would prevent each of these attacks (should it prove pracical)

[0] https://news.ycombinator.com/context?id=39431680

This differs from DRM in a fundamental way, namely that you are not required to do this. You can choose to ignore it if it turns out the object isn't authentic. Whereas DRM relies on making you enforce certain rules, and putting technological or legal barriers in your way to prevent you from breaking them.

But some hard-to-clone and viable-to-read tagging technologies can be very useful for purposes like authenticating genuine product (which consumers are starting to care about), for tracing graymarket items (sold in contract-violating ways, which some international brands care about), and for disrupting retail theft rings.

There's also engagement angles that a lot of brands want to add to the tech investment. (And there may also cost be savings/consolidation opportunities with packaging costs, and it can involve the industrial design, and affect branding, and involve manufacturing, and distribution, and retail, and multiple IT integrations, and field operatives. Which is how your enterprise sales meetings get 10x more stakeholders than the person charged with reducing counterfeiting.)

(I worked on applications of hard-to-clone tags a few years ago. Unfortunately, our startup launched its first factory integration just as Covid hit the US, which meant that investors and customers understandably both got skittish about the near future, we ran out of runway, and our talent scattered to the winds.)

I mean it's basically a hard to replicate barcode.

I still want an authentic pacemaker.

The ability to prove the origin of something over "Source: dude trust me bro" is critical to a great number of people.

But sure, let's stick with "yeah nah, this piece of paper says what's in the box is in the box so it must be legit" because FOSS is gonna get salty about this.

You can hear plenty about the use of xrays (and how unwieldy that is) to verify electronic circuits in some CCC videos. I commented about some of the challenges of false negatives in another version of this thread (which was strangely posted three times by MIT? Is reposting daily until you get traction acceptable behaviour - may have to try it myself some time :)

Well raise my rent.

But I agree it sounds funny. A bit like, "we found faster than electromagnetic waves: we are using light"

But I did notice the pun.

That's a good thing, right[1]?

Something about "the one thing all models have in common..." comes to mind.

[1] https://en.wikipedia.org/wiki/Velocity_factor

But you can't say that blue light travels faster than red light, because that's obviously wrong.