Hacker News new | past | comments | ask | show | jobs | submit login
Postbank to replace 12M bank cards after employees steal 'master key' (timeslive.co.za)
315 points by zdw 19 days ago | hide | past | favorite | 184 comments

Insane. Just think of the risk that this 'master key' exposed to the bank's employees.

Having access to something as insanely valuable as a bank 'master key' puts the employees at risk of blackmail, extortion, etc.

That's why you have HSMs, key ceremonies, Shamir's secret sharing etc. It's not just for trust, it's also for protection of those involved.

Unauthorized wire transfers can be undone, or covered by insurance. Loss of life cannot.

I appreciate this. I wasn't familiar with Adi Shamir's secret sharing scheme. I found this video quite helpful in clarifying how it works for cases where only a subset of the participants are required to confirm the secret.


The original paper on this “How to Share a Secret” by Adi Shamir is incredible. You should absolutely read it.

- The system is proven to be information theoretically secure (not just computationally)

- It uses only high school math

- The paper is only TWO PAGES LONG

I highly recommend printing this out on a single sheet of paper (double sided) and digging in.


Thanks for sharing. This is indeed genius and so simple. Brilliant.

Thank you, I hadn't seen this before and "incredible" is definitely appropriate.

Yesterday there was a Show HN about a purely client-side (?) implementation of Shamir's secret sharing algorithm.



Additional references from the app's About page.

- What is Key Sharding? Shamir’s Secret Sharing Explained (Easy) https://www.youtube-nocookie.com/embed/TQ-DsEZBuQY

- Shamir's Secret Sharing - Solution and alternative to Lagrange (More in-depth) https://www.youtube-nocookie.com/embed/rWPZoz0aux4

Dark Crystal is another neat POC Shamir Secret Sharing tool, built by some folks involved in SSB[0] and Loomio [1]


[0]: https://scuttlebutt.nz/ [1]: https://www.loomio.org/

That was an amazing video. I understand some fundamental crypto about secret sharing with Shamir's algo and it only took 7 minutes. Oh, and in the last 30 seconds they basically explain how bittorrent missing chunks work too. Thanks!

Interesting, thanks for sharing (I too was curious).

This is my favorite video on Diffie-Hellman key exchange which is a slightly different problem (sharing a secret, without the constraint of dividing the key): https://www.youtube.com/watch?v=YEBfamv-_do (oh, hey, same channel! Guess I'll need to check out more videos from here!)

When I was doing IT support one off the trickiest things was trying to get people to stop giving me their passwords. I am not interested in that sort of liability if something gets stolen!

Related to that: why are US landlords so hot on a copy of the house key? In a genuine emergency you let yourself in with an axe, in all other cases it's just a liability with no upside.

It's more a convenience thing for the tenant. Landlord can perform maintenance without the tenant being home, and door guards are common otherwise. Plus US laws tend to give landlords more power.

As a LL I don't want to break down my door to fix a leaking faucet.

It's also so I can enter for regular maintenance without needing the tenant to be there.

HSMs make your storage more secure, but don’t really solve the problem of people having to have access to stuff. You can address those issues, but not fully, and it’s much more difficult. I’ve worked in a number of banks where my admin accounts have given me access to things like SWIFT credentials [0]. I’ve always hated having access to stuff like that, but when you start trying to figure out how to fix it you realize that no matter what you do you’re always going to have important assets protected by secrets, and people are always going to have to have some level of interaction with those secrets.

[0]: https://en.m.wikipedia.org/wiki/Bangladesh_Bank_robbery

Note that this is a banking division of South Africa's Post Office. I was confused it was the Deutsche Postbank.

They lost more than $3.2 million from fraudulent transactions and will now have to replace more than 12 million cards for its customers after employees printed and then stole its master key in December 2018. Took them long enough to figure it out.

>The breach resulted from the printing of the bank's encrypted master key in plain, unencrypted digital language at the Postbank's old data centre in the Pretoria city centre

I can't read anymore. Is there anymore technical explanation?

Was it actually 'printed' ... on paper?

Was it even an actual encryption key or just a password or something?

While the details are sparse, in card payments, you typically use an HSM and provision keys to card blanks using the EMV "personalization" protocol. It derives a bunch of keys from the master secret, and the whole chain of trust is based on these key derivations. These are symmetric keys/secrets, since most chip cards don't have the processing power for RSA, and ECC is still too recent for some of these legacy payment systems.

What it means is that the master secret in the HSM is probably a 128bit AES key. The ways to generate this depend on the ceremony and the particular HSM, but the risk I have encountered in security consulting is the question of whether that master secret itself was just derived from simple components, like pbkdf2("secret phrase known to 3 people", 1000)

The consequence of compromising that key is that an attacker could use it to forge cards, or more usefully, instantiate a virtual card in software to fuzz cryptograms for different account numbers to get available balance information and then personalize cards for those.

People print their bitcoin keys all the time. Sometimes physical security is easier than electronic - current situation not withstanding.

Even KeePass recommends to write the master secret key on a piece of paper. Sure, it must be in a secure place, but we cannot avoid this step for reliability.

No it’s not

An A-Z character string written inside random page a book stored at a relative's house is a super easy way to secure a physical backup of your private key. You could even get clever and encrypt the key using some memorable passage of another book. Since the book passage is unknown to anyone else and be selected from a near infinite pool of paragraphs from the entirety of all published books, it would be impossible for someone to bruteforce even if they found the book where you stored the encrypted key. This would be an easy enough process to generate the ciphertext and store it in the event that you totally lost access to all of your electronics, online accounts, and home.

You are not adding any more security than that if you simply remembering some random key (which isn’t that hard if you put your mind to it).

Don't forget the I and the A in the CIA triad, which are likely to be improved by keeping things recorded in writing.

You could just save it on a flash drive and hide it in safe. How does printing it on a piece of paper make it any safer?

With paper you can be confident that you will still be able to read it in decades. How true is that of a flash drive?

Did you even bother viewing the linked page?

> The advantage is that any security issues of USB interfaces or cameras are completely avoided.

There's also the section that enumerates all the downsides for each medium of storage: https://en.bitcoin.it/wiki/Cold_storage#Private_key_backup_s...

Isn't paper actually one of the more secure storage mediums? If you asked me the best way to store a secret I'd say put it on paper and lock it in a safe. I'd probably do something like print a QR code with an encrypted secret and store the key on a separate printed QR code.

Paper in this context is a form of cold storage, you could get very similar results encoding the information on magnetic tape or on a hard drive stored in a safe. Cold storage turns the information security problem into a physical security problem. The catch comes when you need to use this information regularly, say to issue bank cards. Now you need protocols for regularly circumventing that physical security and correctly using/distributing the information, making it in essence an information security problem again.

A QR code on paper has the benefit that it lasts centuries

It depends on what kinds of attacks you're worried about. One of the challenges with paper is that someone can take a photograph of the data and steal it without the physical secret ever leaving trusted hands.

Probably a backup. It makes sense to have an offline backup in cleartext (for DR), as long as you have the appropriate storage and security controls in place to protect it.

It makes absolutely no sense. Such highly valuable secrets are usually saved using Shamir's secret sharing with parts of the split secret held by people unlikely to collude. Key ceremonies are done in a way that at no point a human being is in position to single-handedly extract the secret from its HSM.

This is a huge failure.

What surprises me is that you can extract the private key at all - I would expect it to be firmly inside a HSM, that only accepts signing requests and the key never leaves the module & the HSM wipes the key if it detects tampering (there is usually a battery inside to power the tampering detection even if the device is not plugged in).

So just exporting the private key so easily without some pretty involved hight-tech HSM key extraction sounds insane.

I worked on a new internal PKI, nothing as high stakes as banking, but we did have a security consultant who had been around the block. We did, of course, purchase such HSMs and design a signing ceremony. But in his opinion it was also normal and expected to keep a decryptable copy of the private key, in case of e.g. changing HSM vendors. It would be even less accessible, but it would exist.

That's a good point - I guess if you know what you are doing then I guess it's fine.

One article said they key was left in clear on a laptop, so maybe a HSM migration was involved.

Interesting.. What Open source tools do you use for this? I would love to read further how I can not have a printed copy of a master key in a safety deposit box.

For card system master keys, you don't use open source tools, you buy a set of properly hardened hardware modules, and follow the appropriate (documented, tested, verified) 'rituals' on them. IIRC the set we used cost something in the ballpark of $50k-$100k, which is not really much compared to all the other things that are table stakes of doing it properly. You can do it much cheaper if the risks are lower and you need less tamper resistance and auditing because you're less likely to have (for example) one of the trusted authorised employees be malicious and willing to invest nontrivial effort in circumventing the system.

Because losing these keys can be very, very, very expensive.

For an example (somewhat similar to this Postbank case) see India Cosmos Bank 2018 incident (https://www.reuters.com/article/cyber-heist-india-idUSL4N1V5... is one link) where criminals generated fake cards to cash out some $13 million; and replacing 12M bank cards also has an huge cost to replace the cards (perhaps roughly $12M - $1 per card replacement is plausible though possibly on the cheap side) even if we ignore the reputation cost.

Here is an implementation of this:


They have a page for each signing ceremony:


With a script of everything done:


You usually use your HSM vendor tooling, in the KCs I was involved the backup secrets were not only split between multiple holders but each holder held its part on a PIN protected smartcard.

We keep printed keys in a safe. I suppose if we were a bank we'd do it differently. They're in "tamper evident" envelopes, too. These are AWS master keys, Bitlocker encryption keys, etc.

HSM's properly implemented do not result in clearext backups of these highly important keys. You can do backups without that.

People seem to forget why credit cards came into existence.

It was not for security. Ever.

Credit Cards were introduced as a less-secure-but-more-convenient-check.

The store then would have a stock of "blank checks" with absolutely no security features where they would imprint with carbon paper and a pressure roll the credit card information and pretty much "mint" the client a check on the spot.

Over time the raised letters for the crude minting press morphed into a magnetic strip, but the process was still 100% the same. Outside of the US in the last decade (2yrs in the US) some little security was added with encryption keys and PINs. Which is nothing more than a digital signature the bank may or may not check (like it did with the actual signature on the previous mentioned blank checks minted by the store). This is the step that was compromised with the stolen keys. In other words, the few places where you have to insert your card chip into a reader and type a pin had their security degraded to the same level as places where you simply use your magnetic strip or type your numbers on an online store.

>Credit Cards were introduced as a less-secure-but-more-convenient-check.

Yeah, no. Diners Club was the first credit card and was released in the 1950's to aggregate and streamline paying for things on account. It was, as the name suggests a line of credit. Checking accounts could become credit accounts if they allow overdrafts but this is not the normal intent.

It was not until First National of Seattle released a debit card (also now known as a check card) in 1978 that any of those plastic cards behaved as a check.

It's an important distinction because in one case (credit) the money transferred first is the bank's. This has important implications on who holds the liability and how long investigstions take into fraud allegations.

Chip security was added first overseas because online authorization was less ubiquitous. The US was able to have online terminals pretty much everywhere since the 1980's and knucklebusters became a rare sight in the 1990's. The chip security now is less to prevent card-present fraud in the US, more to prevent reauthorizations of stored CC information (see, target breach). With a chip or NFC the card generates a unique signature for every transaction so there's nothing a POS system could store to reauthorize future transactions.

This is why in the US for credit cards a pin is mostly unheard of and we are finally moving away from signatures which have been obsolete since the time offline batching mostly went away.

Edit: Consider 1980's America was under the monopoly of AT&T. This greatly influenced the design of payment card networks. Europe, for example, probably had probably over 100 different companies in dozens of countries, so universal online authorization was less feasible.

You are completely missing the point by an internal bank semantic.

Credit cards, yes, started replacing the high APR credit-checks. And then moved to a mix of this and regular debit checks.

But the actual process, which was what i was describing, was still pretty much exactly the same used to validate a check (whichever type) at the bank branch. Where the bank will draw the funds from is not a important topic when we are talking about process of the transaction itself.

The few places is everywhere in the EU at least. Signing card transactions is very last century in more advanced nations. ;)

Edit: also the pin is verified each Transaction unlike with receipt signatures, which as you say were rarely checked by anyone.

>also the pin is verified each Transaction unlike with receipt signatures

At least in my experience this has increased a great deal more in the US in recent years.

Also, nearly everywhere in Canada, the interac system really has it together up here.

It helps having one debit card system to support. The US has like 7 and I couldn't begin to name most of them (Interlink is one of them, I think Maestro is another)

I don't really buy that though. The banks may have an easy time talking to other Canadian banks but they also need to talk with American banks and Canadian PoS terminals often support US debit cards. Perhaps Euro PoS terminals just refuse to work with US debit cards but Canada gets a lot of those American tourist dollars and doesn't want to make it hard for you to spend your money up here.

> Canadian PoS terminals often support US debit cards.

Unless something changed in the past few years, this is not the case. US debit cards are accepted through Visa or Mastercard's payment network which doesn't require a PIN (this is what it means when you process as credit in the US). Those transactions do not use the debit payment network. https://en.wikipedia.org/wiki/Interlink_(interbank_network).

EDIT: I will add that while POS systems don't have a way to do this up there, ATM's with their notoriously high fees do support cash access through a US debit card.

Maybe 5 years ago or so a store in London pulled out their carbon copy machine from under the counter to accept my American credit card.

Happened to my wife about 10 months ago when there was a power failure in a department store while she was in the middle of a transaction.

I've seen them in taxis once in a blue moon if their card reader is broken.

Even more fun fact: we rely on mag stripes (vs. chip n pin / nfc / etc) because of gas pumps. The cost of refitting gas pumps holds us back. Yay.

Liability shift for gas pumps has been pushed back (October 2020 I think) but not eliminated.

Liability shift works like this: Historically the merchant is protected if they had good faith belief that the payment card was authorised via things like a mag stripe. After liability shift that goes away unless you use EMV. You checked the card was "real" with just a mag stripe? Not interested, the chargeback comes out of your money.

Your local gas station can keep using mag stripe readers for another decade if they like. But liability shift means places that see significant fraud will have an obvious economic incentive to go EMV, and that shifts the fraud onto nearby stations that didn't have fraud problems, so they go EMV and so on.

If you're a merchant with very low fraud rates it can make sense to do no Authorization step whatsoever. Any merchant, anywhere in the world, can do Settlement, which is the step that moves money from your account to theirs, based just on the card number. Only if it becomes a question as to whether this payment was authorized does it matter whether they did the Authorization step, a mag stripe read, the old fashioned impression machines, or a chip-and-PIN terminal.

Certain types of companies can find it makes sense to do no Authorization for groups of customers. One Click may be an example of that. You did a bunch of transactions, with a physical delivery address, what are the chances that a new transaction with the same delivery address and card details is fraud while the previous ones were not?

Liability shift or not some of these gas stations keep insisting on ID with credit cards.

I am not sure that is the problem. Timing was a big problem.

When NFC showed up, it was intended for plastic cards. This was pretty widely deployed. Then software companies integrated it with phones. This made the telcos unhappy, because these phones had a "secure element" that they did not control (traditionally the SIM card was the secure element, but these phones ignored that and that upset them; back then, a carrier being upset meant that your phone could not be used on their network). It also made processing networks unhappy, because they saw that they were losing control. (You don't need Mastercard and Visa when the phone can just use the Internet to ask Apple to authorize the transaction, after all.) So they flat-out stopped issuing cards with NFC. Then the final blow is that merchants were tired of paying credit card transaction fees, so they removed NFC readers from their stores, and banded together to make some shitty system to bill your purchase directly to your checking account. No more paying fees or pesky chargebacks.

The retail side blew up -- no consumer wanted it, and it was technologically bad. NFC readers are back in stores. The carrier side blew up -- SIM cards are gone and Apple or Samsung is your secure element provider. I am not sure what happened on the processing card network. I'm guessing they made some private deal with the NFC payment providers (Apple, Google, Garmin, Fitbit, etc.; enough companies in on the game that they can sit back and watch them fight each other while they profit).

I kind of got to watch this from both sides. I worked on Google Wallet when I started at Google in 2012. Used it pretty much every time I went to CVS. Then CVS removed their readers. Then Apple entered the market, and CVS once again accepts NFC cards.

So maybe gas pumps played a role; I don't drive so I don't interact with gas pumps ever. But there were much deeper problems. A lot of entrenched monopolies stood to lose a lot, so they were happy to impede progress wherever possible. There was never a good possible outcome, though -- let the incumbents keep their power, or let the upstart megacorps become the new incumbent. Plenty of VC money available if you think you can fix this problem, or become the new big guy ;)

>The retail side blew up -- no consumer wanted it, and it was technologically bad.

Didn't help that thr superior user experience of NFC built into cards was drowned out by the klaxons of the media continuously warning customers of proximity theft.

Wrong threat model, the cheap mag skimmer or camera or unsecured database models were the real risks. It wasn't that someone will stand butt-to-butt with you to steal a NFC token.

There isn't anything to really "steal", a contactless card is willing to participate in transactions but "I am standing next to you" isn't a transaction on its own.

You need to do a relay attack. Here's how that goes:

1. Jenny's payment card is in her jacket pocket.

2. Charlie walks into a store wearing a small NFC-capable computer and a medium distance radio (a cell phone might do) perhaps concealed inside his clothing

3. Charlie's friend Barry walks near Jenny, Barry is also wearing a similar setup to Charlie.

4. As Barry gets close to Jenny, Charlie "checks out" at the store, paying with NFC. The transaction travels from a machine near Charlie, through the radio, to Barry (now creepily close to Jenny) and then back over NFC to Jenny's card. Jenny's card agrees to the purchase - everything seems legit.

Jenny just paid for Charlie's purchases even though they've never met.

This attack isn't economically attractive because transaction sizes are limited. A complicated trick that sometimes allows you to get "free" pizza or coffee in exchange for risking time behind bars seems like a bad idea. If you could get a laptop, or a big TV then it might be more attractive, but you can't because those cost too much to allow mere NFC presence authorization.

What about mobile payment terminals (where I live, they're very common)? Instead of virtually moving the card to the store through a relay attack, move the store to the card through one of the spare payment terminals (many stores have several as a backup for when the POS or the network is down, or for deliveries) hidden in the clothing. Of course, this changes the threat model a bit, since it now needs collusion from a store employee.

Let's suppose that you take a mobile payment terminal, get on the train, and make ten $10 charges (you can't do large transactions contactless) through clothing of some people. What does that mean?

First, it does not mean that you're getting any money. Such a charge is effectively "sending an invoice" to the issuing bank from the store which is supposed to have that terminal, and they will pay your merchant bank, which will give money to the institution who got issued that terminal. There's no way for the store clerk personally or someone else to get to these funds.

Second, the money is not coming today. You get an authorization message, but you'll receive the actual money later.. if ever. If the payment is disputed, you won't get that money. If it's disputed a month later, they'll take that money back from you. If many of your payments are disputed, then all your incoming funds will be frozen until they verify if all of them should be returned. If multiple payments are disputed, then the standard methods of tracing 'common point of purchase' will reveal the particular terminal as the culprit. Also, malicious merchants is a known threat, so the merchant bank will ensure that you can't just spam a day's worth of fraudulent purchases and run - standard terms will expect that some amount of money is frozen (e.g. rolling 15 days worth of transactions) so if you suddenly get a bunch of chargebacks, the customers will be paid back in full. Fake stores and shell companies are a thing, but there are reasonably effective measures to try and prevent that.

So there's no threat through collusion from a store employee - the fraudsters would get identified and would not get any money at all; and there's limited threat from collusion with a whole merchant - the fraudsters would get identified and can't get any meaningful amount of money. Extracting a couple hundred dollars could probably work - but you're "burning" the identities of multiple people and a company; the bank will 'eat' that loss if you succeed, but you can't repeat this trick.

> Of course, this changes the threat model a bit, since it now needs collusion from a store employee.

An employee? Or the store owner?

These terminals don't offer a choice of destination bank account, if you type $16.94 into Walmart terminal and hold it near Jenny's card the card will authorize $16.94 payment... to Walmart.

For an independent merchant (including e.g. franchise operators) in some sense that's their money in the merchant account, so it makes a little bit more sense, but I still don't think it really adds up. It's like opening a bar so you can get cheap booze, the economics don't make sense.

There are opportunities for insider crooks. In the UK for example there were a rash of what are morally skimmers built into chip card terminals. Here's how that worked:

You own one or more stores with shiny new EMV payment terminals. From an instructional video you learn how to prise open a common model of terminal without setting off its tamper detection. Then you use the huge space inside the terminal left for an optional security feature (never implemented because features costs money) to add a board that taps the communication to the card and uses a cell phone connection to upload it. You seal up the tampered terminal and install it at one of your busier stores.

Customer puts their chip card in, the terminal works as expected but unknown to them your modification stores the card details and transmits them to other crooks half way around the world.

The other crooks are making old-fashioned magnetic stripe cards with details that have been uploaded. They send small fry out with these bogus cards to buy stuff in a country that doesn't have EMV yet. The stuff is fenced, and you, back in the country with EMV, get say 10% of the proceeds for your contribution to this international crime.

Some people in the UK got prison time for this. International card fraud is easier to spot (this person bought groceries just outside Luton, then forty minutes later they bought a laptop computer in Hyderabad or Houston?) but until EMV is rolled out everywhere similar tricks will be done.

So why not limit the magstripe to gas pumps and give them a time frame of 5-10 years to upgrade existing pumps, and mandate chip capable card readers in new pumps?

At the same time fees could be raised 2% for non-chip transactions to incentivize upgrades.

Looks like they have deadline of 10/2020 before visa and mastercard start holding owners liable for fraud if they haven't upgraded to chip readers.


That's stupid. Just issue cards without stripe and the problem will solve itself.

It's a chicken and egg problem. Gas purchases are on the of the most common purchases in the US. Removing the ability to pay using a credit card would a) piss of owners of the new credit cards, b) likely cause people to use their old credit cards for longer and still wouldn't push gas stations to switch over quickly as most have atms and gas stations could insist on cash payments. Keep in mind that at least some of these gas pumps can't be upgraded but have to be replaced in order to support the new chip cards. What should have been done is a combination of good incentives for upgrading early (maybe reduced fees?) and penalties for not upgrading within a time frame (fraud payments are owners responsibility after n amount of time, increased fees for dealing with fraud)

That's stupid. Just issue cards without stripe and the problem will solve itself.

I don't see a bank replacing its customers' forms of payment with a new form that isn't accepted in as many places as the old one.

We already have incentives in place in the US to stop using mag stripe.

Shops accepting mag stripe payments are liable for counterfeit fraud. If you use EMV (chip cards), then card brands protect you (Visa, MasterCard, etc.

You can Google it as "EMV liability shift" if you're interested in more details.

I'm starting to see chip enabled pumps. Unfortunately they aren't obviously labeled so you assume it needs a swipe and have to wait to restart when it complains about not reading the chip.

At the chip gas pumps, I'm seeing paper signs left by the owner that say "Leave card inserted". I'm glad everyone is going the chip route but I still wish that pins would be required for credit cards. Just having a 4 digit pin is loads more secure than expecting some cashier to verify a signature on the back of the card (which never happens). I can still bypass the pin on my debit card in most situations by just entering in nothing unless I'm doing cash back. I'm not really sure how that's supposed to protect me, whats the point of a pin if you don't need to type it in? Card companies are quick to refund you for any fraud but it creates an inconvenience for the user that really just isn't necessary. I've been on vacation abroad when my credit card was used fraudulently which I had to cancel immediately. Luckily I had access to my corporate credit card I could use to pay my hotel bill since that was the only thing I couldn't afford with the cash I had on me but I very well could have been shit out of luck. I paid the charges off myself and wasn't in too much trouble after I explained the situation.

I'm starting to see chip enabled pumps.

Exxon (pre-ExxonMobil) had chip-enabled pumps in the early 90's. Instead of a card, you had a little cylinder-shaped keyfob about the size of a few Tylenols strung together. I'm not sure what happened to that, I ended up moving out of an area served by Exxon.

These days I see gas pumps with the little wireless payment logo, and I have a card with the same logo, but it's never worked for me.

How else would you store it? Printed is not vulnerable to electronic hacking, and encrypting it would be pointless since you’d also need to store the encryption key to the master encryption key then.

Ah hah hah haha!

And our (USA) law enforcement agencies promise us that any encryption master keys required by their grandiose plans will only be used in cases with proper legal court warrants (ignore the FISA court warrant abuse based on lies and deceit) and will be super secure and never stolen.

Just like those secret hacking tools stolen from the CIA.

Or these private master keys.

Don't forget the TSA travel master keys, which can now be 3-D printed by anyone using this repo: https://github.com/Xyl2k/TSA-Travel-Sentry-master-keys

You can use a ball point pen to pop open the zipper on any luggage.

I understand that the illusion of control is very helpful for nervous passengers, but your luggage is leaving your control and it's mostly nylon fabric and plastic.

Yes, but you're missing the forest for the trees. A government entity (TSA) had a thing built specifically for their needs ("secure" locks for luggage) and promised they would be the only ones capable of unlocking it. Then somehow the "secret" they promised to protect (physical keys) got leaked out somehow and now the entire thing is security theater.

Also I think they used these locks on handgun / firearm containers that were declared when travelers with a CPL or LEO people traveled. Those are typically put into a hard container that is difficult to open unless you unlock the lock.

EDIT: last paragraph is wrong, please ignore that. Thanks guys.

I don’t think you have to use a TSA lock on luggage with a firearm [1]. If it meets the legal definition of a “firearm” you trigger (get it?) the TSA rules that allow you to use a non-joke lock. Since this includes just the legal firearm you don’t even have to carry an actual gun to prevent the TSA from sniffing your undies.

Deviant Ollam gave an (in)famous talk [2] about this at Defcon.

[1]: https://www.tsa.gov/travel/transporting-firearms-and-ammunit...

[2]: https://youtu.be/KfqtYfaILHw

That's a funny hack. However, this looks like conflicting advice:

"Only the passenger should retain the key or combination to the lock unless TSA personnel request the key to open the firearm container to ensure compliance with TSA regulations. You may use any brand or type of lock to secure your firearm case, including TSA-recognized locks."

If you use a TSA lock, that means they have a key to open up the case and access the firearm without you around. That's a big no-no and could result in problems with the law.

You'd also probably want to show up an extra hour early because this could cause long delays. Some airports are notorious for workers who don't know the regulations, don't want to be responsible for any of it, and will refuse to do anything to help you resolve the situation. Also, definitely don't try this if you're flying through New York or New Jersey.

I agree that paragraph seems contradictory. I guess I have a slightly lower opinion of the TSA now.

What’s important is that you don’t have to use a TSA lock. You just have to open it for them if they ask.

> Also, definitely don't try this if you're flying through New York or New Jersey.


CPL's are typically granted at the state level, but not all states need to recognize your CPL as valid. So it's perfectly legal for me to fly from Detroit to New York with a firearm, because the TSA staff in Detroit, MI legally recognize my MI CPL.

However, the TSA agents in NY do not recognize my MI CPL. As a result when I fly back from NY to MI, I could get arrested for carrying a firearm.

More on this here: https://www.theblaze.com/news/2018/01/18/colorado-woman-decl...

That has to do with concealed carry. Putting a firearm in your suitcase does not require a CPL, nor does it require concealed or even open carry. I'm sure it's still legal to transport a firearm in New York.

To be clear, this rule can be triggered by a flare gun or starting pistol. Traveling with those seems perfectly reasonable and I don't see why a CPL would be required.

> If you use a TSA lock, that means they have a key to open up the case and access the firearm without you around. That's a big no-no and could result in problems with the law.

Why? If a TSA person unlocks it in accordance with their procedures, surely that's fine (and is what they would do anyway, just getting the key from you first)? If a TSA person abuses their position to open your luggage with their master key against their procedures, surely that's not your problem - more akin to a criminal breaking open your luggage than anything else.

The problem is not access to the luggage. The problem is access to the firearm. That is why when you are transporting a firearm TSA does not retain the ability to access the firearm outside of your supervision.

TSA's procedures are as robust as a high school recital of A Midsummer Night’s Dream. The TSA knows this and so when it matters (like when firearms are involved) the rules change.

Most locks have master keys and most find their way into the hands of sufficiently motivated criminals, right? Is there any lock you could put on your firearm case for which there wouldn't be authorised people who had the master key to it?

I don't think your initial assertion is correct and even if it is nothing requires you to use such a lock.

Most high security locks can be re-keyed. There's no need for it to be keyed to any existing master. I consider that better than "one of these 6 predefined TSA keys that can be 3d printed oh and also it is labelled."

Nor does the threat model of "dedicated attacker" make a lot of sense. If that is your constraint then flying with complete guns is probably not for you.

There are easier ways for criminals to get guns than a mole in the TSA or baggage handling. Like, stores. A decent padlock can secure a gun case beyond the "crime of opportunity" level required for protection in a secured environment like an airport.

If you're really paranoid you can even build in trap pins so attempting picking or using the wrong key can permanently bind the lock. The point is you are in full control of the lock and don't have a government mandated, well understood backdoor included.

Finally, all of this misses the point which is that you can put a firearm such as a starting pistol, flare gun, frame or lower [1] in your luggage and use whatever lock you want. This means you can protect the rest of your stuff from getting swiped by TSA or baggage handlers.

[1]: https://www.atf.gov/firearms/firearms-guides-importation-ver...

Can you explain how using a TSA lock on a gun case you're flying with, if you want to, is "a big no-no and could result in problems with the law"? That's the part I'm disagreeing with. Surely if a knave uses a stolen master key (or makes unauthorised use of a master key they have as part of their job) to open your gun case, that's not your liability.

> Can you explain how using a TSA lock on a gun case you're flying with, if you want to, is "a big no-no and could result in problems with the law"?

I'm not sure, those are not my words. Ask swimfar.

It strikes me as simply irresponsible gun ownership although I'm not sure what legal liability that may open you up to.

Again, the point of bringing this up is that you can decide to use an actual secure lock to protect your property if you just put a "firearm" in your luggage. It's not actually about firearms at all. It's about protecting yourself from an obviously broken and exploitable government backdoor.

> and now the entire thing is security theater.

I am not the one missing the forest for the trees.

It has always been security theater. From day 1. What's most galling is that quite a lot of the people participating here are young enough that they don't remember a time before the theater. If you are under 30 your first plane flight you can remember is probably after this all started.

Locks are mechanical, and a master key can be built by taking apart a lock and measuring the innards. In some cases you can create a master key by filing down the right teeth on a key you already have. Master keys in apartment buildings are an old, known problem to competitive lockpickers.

As other people are pointing out elsewhere in this thread, carrying a firearm requires that you use a lock-lock, not a TSA "lock".

> got leaked out somehow

They let the Washington Post take high-res pictures of the masters while supporting them doing a story about the TSA and those got published which was enough to decode the keys.

Yep. And then you can just run the locked zipper back and forth over the opened section to close it, without any evidence of it having been opened.

Simply put: putting anything you can't replace into checked luggage is foolish.

The TSA is a lot more likely to steal your stuff than some random person with a printed key. Plus, if you use a real lock, they have the right to clip it off, which is trivial. Don't put stuff of value into checked luggage. Keep it on your person or ship it via a carrier who has insurance.

> Plus, if you use a real lock, they have the right to clip it off, which is trivial.

Life hack: you can pack a stripped AR lower, and it's legally a firearm. You must use a non-TSA lock, and the TSA (nominally) is not allowed to open it.

Though your core point is valid: your nylon luggage isn't going to keep anyone determined out.

I did not understand, what is a "stripped AR lower"?

And why is TSA not allowed to open it?

Edit: I guess AR is Assault Rifle?

In the US (not sure about your location or the laws in other countries), the part of an AR-15 that qualifies as the "firearm" in the eyes of the law is the lower receiver. It's a machined block of metal and does not do anything on its own.

When transporting a firearm through air travel, it's supposed to be locked such that only the owner can open it. So you can claim you're carrying a firearm with nothing but a small-ish piece of metal in your luggage.

For illustration: https://i.imgur.com/ARZgZV6.png. The non-transparent part is the lower receiver.

As an aside, the "AR" in AR-15 stands for Armalite Rifle (after the company that originally designed it); not "assault rifle".

thanks for the explanation. I am not living in US, so not very familiar.

Or check a metal case with a lock the TSA can't open: buy a flare gun (legally a firearm, cheaper than normal guns), declare it, and have your other valuables in the same bag. Legally firearms must be in checked luggage, without ammunition, in hard cases (not soft bags), locked by a mechanism only the owner can open.

But unless you travel with valuables quite often it is definitely cheaper to ship it insured.

This sounds like a better idea than flying with AR-15 parts. Less likely to scare people, and less likely to raise flags or get put on some list.

I feel like a flare gun is more likely to scare than a lower receiver. A flare gun immediately looks like "a gun" whereas a block of metal doesn't look very threatening (assuming you don't have a tacti-cool one with skulls and garbage on it).

I only flew with a stripped lower once, and the baggage attendant at the Delta counter had to wave a TSA agent over because they did not believe it was a firearm. I never had an issue checking a handgun though.

More likely to scare, but since it's in a sealed case and looks like a firearm it's easier to get the attendant to accept that it is a firearm when checking it. It matches their expectations of what a gun should look like. And since it's in a sealed case nobody unexpected will see it anyway, so scares are unlikely.

Finding a smallish shoulder bag that I actually loved instead of just tolerated went a long way toward helping me winnow down my traveling kit. You trade a little more up-front hassle at the X-ray machines for keeping an eye on your stuff and not having to bother with the baggage carousel.

Those things were useless anyway, based on my experience. The one time I used them, TSA neglected to even try opening the lock carefully and simply broke the zipper getting into my luggage.

> Recently, pictures of TSA master baggage keys got leaked by the Washington Post and also PDFs hosted on TravelSentry's Website.

I'm glad we don't need to trust the TSA with following any kind of security protocols. /s

Postbank is a subsidiary of South Africa's Post Office.

An entity that is bankrupt and barely functional despite having a state mandated monopoly on an entire country's postal system.

As per article they're also running the SASSA social grant system which is a train wreck in itself and has been buried in legal disputes for years (not random small cases...a challenge to the legitimacy of their core mandate on grants).

Someone walking out the door with printed encryption keys sounds about right.

> An entity that is bankrupt and barely functional despite having a state mandated monopoly on an entire country's postal system.

To be fair, plenty of state-mandated monopolies of postal systems still aren't profitable, often because they don't significantly control their revenue stream insofar as they don't set the prices of their products.

Your argument is broadly sound, but not applicable in this particular case.

Currently people are paying a massive premium to utilize private courier to have their stuff privately couriered. People are literally going f that I don't care what it costs I don't trust the postal system.

A sane postal service would review this situation and try to understand why the people think "f you". SAPO has decided that the correct solution to this situation is to simply legally mandate that everyone has to use their useless service thus making couriers illegal for this category. Literally:


I personally take a pretty dim view on "nobody wants this...fine...so we'll just force it by law" dynamics

Unfortunately, I don't think there is a single state-run company in South Africa that isn't bankrupt or on the brink of bankruptcy.

The key factor has been unmitigated corruption and ineptitude.

To clarify: no one is compelled to use the South African Post Office. There are dozens of courier companies that offer much better service (legally). The SAPO operating model is the same as many/most other legacy postal services around the world, but more incompetently executed.

>no one is compelled to use the South African Post Office.

They are actually - and SAPO is trying to legally enforce it. Whether it sticks or not...we shall see.


That's not true. When you order something that goes through customs, it will be sent to a random SAPO (not even necessarily your closest one or the same one each time) for you to pay the customs fee in person with cash and sometime card. This is regardless of whether that package was headed for processing and distribution by one of the private post offices.

It seems ridiculously cheap, relative to payout, to attack financial institutions by joining them as employees. Couple this with the fact that "white collar" crimes are punished with relatively lenient sentences.

When IT leadership talks a big game about security, as a skilled practitioner on the ground I still see plenty of opportunities. When you delve into details, a lot of that “serious business regulated entity with compliance auditing requirements” stuff is brain dead. Like, running vulnerability scanners that don’t even speak the same wire protocols as a the applications under test. Enforcing strict ACLs on UIs while it’s trivial to SSH and curl the backend APIs.

I sometimes wonder whether accounting is like this too. Is financial audit just as much of a fig leaf? Can a skilled accountant also spot dozens of ways to embezzle?

We are truly saved by our better angels. If all people were ruthless economic competitors, and took every efficient financial opportunity available to them, criminal or legal, a cooperative society would be unmaintainable. We should consider ourselves lucky (perhaps blessed) that the anti-social among us are a tail end minority.

This is quite surprising to me. Some of the banking technology and practices in Germany are extremely tight and their infrastructure seems problematic. Living there for a while as a non-native and being with PostBank, it was insanely difficult to login and understand. Germany is also heavily marred with bank cash groups, meaning you get charged elsewhere.

It was great to see N26 come along and change that, and thank god for their native translations.

That's about South Africa's Postbank, an entity unrelated to the German bank of the same name.

They really should add "[South Africa]" the title, since my first thought was "Is this Germany or might they mean PostFinance in Switzerland?"...

How do you summmon dang, with @dang?

I think burning sage and eye of newt is involved too.

I think the German and the South African Postbanks are not related.

Shouldn't this be the kind of key/info/data where it is divided into N pieces, and all of the pieces are stored separately?

Shamir Secret Splitting divides a secret into M pieces requiring N parts to reconstitute. This does not absolutely prevent this attack, but requires collusion between attackers who are presumably trusted insiders, but trusted insiders can be compromised by blackmail, coersion, greed, or other angles.

My accounting professor used to say the mirrors and security cameras in stores were to monitor the employees more than the customers.

That only would make sense if they don’t have cameras in the stock room and loading docks, etc.

I've worked at more than one location where the Cameras in customer areas were conspicuous but the cameras in the employee area were hidden.

Most thefts by customers are more opportunistic, the visible cameras have much more effect on that. Staff theft tends to be more pre-planned when it happens, so they'll work around the cameras if they easily know what they can see. Same for break-ins to employee-only areas (warehouse etc.).

The threat model in the two areas is a bit different, so different tactics are used. It doesn't mean that some people are trusted more than others, just that as groups they behave differently.

Sure but stock room knows the security people and if the manager is causing shrinkage then the staff might also follow “the lead”.

> but the cameras in the employee area were hidden.

Is this legal?

Sounds like a lawsuit waiting to happen.

In the UK at least, if you believe that crimes are occurring (e.g. stock discrepancies suggest employees are stealing) you can get permission from your local police to use hidden cameras.

With that prior police authorisation it's legal to record more or less anywhere without those recorded knowing.

The police knowing you were authorised means when Sandra finds a camera in the employee locker room and freaks out, the cops who attend can verify that yes, Sandra's manager's boss got that OK'd two weeks ago when six £700 graphics cards "vanished" versus no this is some camera planted by an unknown creep and if the cops figure out who it was they'll be arrested.

If the "unknown creep" turns out to be the boss, the existence of this authorisation mechanism means they have no excuse - if they'd asked "Hey, can I secretly film my employees to get evidence of theft?" the answer would be "First, go to a police station to get guidance" not "Sure, act exactly like a pervert and just hide a camera".

What law would you think is being broken here? I don't think they are talking about bathrooms, where one might argue there is a "reasonable expectation of privacy." If you are working for an employer on their premises I think they have a right to do monitoring including clandestine. Just like they can monitor anything you do on their computer

I know several jurisdictions where recording employees place of work is illegal.

Like you can record corridors and such, but you can't record him programming at his desk.

Not in the EU.

Wage theft is hard to monitor with cameras and mirrors.

How exactly would one use this key nefariously? It seems to do so would require a whole chain of capabilities.

In any case it would make sense to have multiple master keys...say use a different master key for each batch of so many cards.. then if one key was compromised it wouldn't affect 12M cards.

I have this fantasy that there was some security engineer who said, "hrm this looks like you could steal/reconstruct this master key by..." raised it up to management and subsequently ignored or shot down because changing the solution was too expensive.

There's a non-paywall article on the subject at:


That proves my point again that there are not enough regulations on electronic security standards that applies to private companies.

All you have are white hat security consultant experts that only have their dollars and reputation to work with. The public is highly vulnerable on those things yet I don't see politicians really caring.

Hardly a private company: PostBank is the banking division of the SA Post Office, and firmly a government-run entity.

And not the first government run entity in SA to have funds stolen from it.

Transnet/PRASA (railways)

Eskom (electricity)

SAA (airline)

All have had just totally monumental theft often at the top levels. It all seems to be pretty consequence free. Meanwhile, the hawks open immediate investigations and charge people almost immediately who are trying to expose things - which is kind of funny - 10 years to investigate totally obvious theft, 7 days to charge folks who are exposing it :)!

South Africa has insane levels of graft and corruption that have been going on for decades. They also have a lot of politically motivated assassinations. Total basket case of a country.

Not really a total basket case. They had a long and fantastic history actually of institution building, including credible tax services etc.

It was really the Zuma years that messed things up and the new folks don't seem interested in any cleanup (the ANC has had a lock on power a long time). They are still a great country relatively speaking in Africa, with incredible potential.

As a US expat living in Joburg, I agree that there is a lot of corruption in SA. But damn if 2020 hasn't shown a spotlight on the shady business in the US as well.

1. Most of the graft and corruption has been going on at least 7 decades. i.e. Long before the ANC government.

2. "politically motivated assasinations"? Citation wanted, please.


> These prizes seem to be worth killing for. Since 2014 more than 115 Glebelands residents have been murdered. Many were anc members who objected to the ways of Robert Mzobe, an anc councillor accused of corruption, and Bongani Hlope, a local warlord who terrorised residents. “Glebelands is a microcosm,” argues Mary de Haas, a researcher into local violence. Throughout the country violence is regularly meted out by one faction of the anc against another. From 2000 to 2017 nearly 300 political assassinations have been recorded, many of them anc members.

Replace "south africa" with the USA and your statement is equally verifiable and accurate.

You are sorely, sorely mistaken.

Quite the opposite: the free market will deal with this just fine, giving a big penalty to companies that don't care enough. The government, on the other hand, imposes bad businesses, enables regulatory capture and has proven many times that it has no idea how to handle infosec. These white hat consultants aren't perfect but through competition they're still better than lobbied lawmakers.

The FDA made it mandatory for food manufacturers to list their ingredients on food they sell. Prior to these laws no food manufacturer was willing to do this freely, for fear of someone else stealing their recipe. But now consumers can make informed decisions and avoid foods that will give them severe, life threatening allergic reactions.

Every citizen who drives a motor vehicle must be issued a driver's licence by the state they reside in. This enables a minimum understanding of how a complex machine (the highway transportation system) works so that other drivers / participants of that system can use it safely. Without regulation, vehicle collisions would be rampant and the system would be far less stable.

Not all government regulation and oversight is bad.

I highly recommend reading the book "Click here to kill everybody" by Bruce Schneier [1]. Bruce has been in the cyber-security world for a long time now and is considered one of the world's foremost experts on cryptography and computer security. In this book he makes projections about the "internet of things (internet+)" and the security problems that will come with it. He strongly advocates for government oversight and legal reform in multiple areas.

[1]: https://www.schneier.com/books/click_here/

> The FDA made it mandatory for food manufacturers to list their ingredients on food they sell

FDA still allows companies to hide ingredients under the guise of "natural flavors" and "artificial flavors". I hope you aren't allergic to one of those!

> Every citizen who drives a motor vehicle must be issued a driver's licence by the state they reside in. This enables a minimum understanding of how a complex machine (the highway transportation system) works so that other drivers / participants of that system can use it safely.

License plate technology helps the government (and private companies) know where-ever you are at all times! https://www.bloomberg.com/news/articles/2018-12-06/why-priva... https://www.cnet.com/news/this-company-could-turn-every-home...

Free markets don't exist outside of textbooks and can't due to the complexities of social relations we have to deal with in the real world. It seems like you're suggesting a solution that can't ever work by merit of it being unable to exist.

Do you think that the free market worked as desired in the case of Equifax?

Do you think we actually have a free market?

If we don't have a free market, it doesn't do much good to say the free market will solve any of our problems.

Huh? Presumably the suggestion is to make the market freer.

Or perhaps to acknowledge that the assumptions about a free-market ignore the realities of human psychology, and thus a free-market inherently cannot exist because humans are not capable of producing a free market without regulation, and regulation is antithetical to a truly free market?

The idea is not that the free market will always produce perfect results and perfect goods. The idea is that the free market will produce better results at a faster pace with less unintended consequences than government intervention.

Government intervention rarely happens proactively, just as major market changes rarely happen proactively. Supplier failure on a large scale makes it apparent that change is necessary, and only then do changes follow, either from the top down by government intervention, or from the bottom up by the market.

Free markets do not capture all externalities.

Humans are a real security risk.

Is there no equivalent of PCI in South Africa?

and why was there a "master" key?

Looks like it's a key related to South african social security agency and cards are replaced because the dudes rampaged the entire bank system.

They probably rolled their own encryption, because it was over 20 years ago.

Would you build a castle with a back-door?

Actually, castles almost always had back doors.


Almost always? Source for that? Very interesting, I guess during a siege there's a tradeoff there. If your postern is on the other side of a mountain range, then definitely it's not so bad.

Do you mean the “main key”?

Does anyone have a link without a paywall?


It's a general purpose publication. The general public knows what a "master key" is. "Main key" is meaningless.

I believe the parent is referencing the github "main" vs "master" controversy, where "master" was considered "exclusionary", and should therefore be replaced with "main".

I thought the problem was master/slave terminology, in say replication.

It started that way. Now any use of "master" is considered problematic by a small group. Same with blacklist and whitelist.

Yes, it was supposed to be a joke about this.

I'm fairly certain this was a joke about the debate going on regarding git and the broader development community trying to be more inclusive by changing the name of the default branch in git. Which is typically called master.

I'm a little bit confused by that whole discussion. Keys are not people. Nor are branches. Bank cards. Hard drives. It seems perfectly fine to have a master/slave relationship between components. I don't quite get how that impacts anyone. Intelligent lifeforms obviously shouldn't be treated as slaves, but that doesn't seem to be what this is about; merely a matter of terminology?

Painters can be masters, carpenters and all kinds of other artists and craftsmen too. Are they no longer masters of their crafts because the word is somehow "not inclusive"? It's not. And it shouldn't be. It takes word to be a master crafter.

What about BDSM, should subs and doms no longer refer to their counterparts as master/mistress & slave?

Apparently this is a controversial issue in some places, and I don't understand why. Maybe HN isn't the best place to gain that understanding, but then maybe I'll be pleasantly surprised?

It's just a lot of pointless outrage about nothing by a couple of people who need to feel they're accomplishing something. Sadly companies seem to follow them as to avoid being targeted by social media 'shitstorms'.

I was hoping that as a non-native speaker I might have missed some point of nuance somewhere.

I'm not a native speaker either. Sadly, this ideology exists.

Yes, it was.

*main key

If they used main key this wouldn't be a problem

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact