3 ms·

Well, as a guy who spend couple of years actually implementing chip & pin terminal credit card application from scratch, for a Polish acquirer, and a security o

Well, as a guy who spend couple of years actually implementing chip & pin terminal credit card application from scratch, for a Polish acquirer, and a security officer at the same company, who had to read and internalize 17 thousand pages of documentation from Visa, Mastercard, PCI and EMV to do so, I can confidently say this is complete and utter nonsense (and it already was 10 years ago when I have stopped working with credit cards). At every step Visa and MasterCard promotes highest possible level of security. To give you an example, when implementing the terminal application you have to be able to support a range of CVMs (Cardholder Verification Methods) that might be presented by the card. The CVM list is a detail personalized on the card by the Bank, issuer of the card. The algorithm actively promotes selecting the strongest authentication method that is supported by the card and always prefers PIN over signature (ie if PIN CVM is available it should not be possible to normally use signature). This is verified during certification process and applications that don't meet this requirement fail certification. There are also rules and internal memos which are sent to both acquirers and issuers, to which I have been privy to, which exert pressure on issuers to improve their security to deal with frauds. We were also regularly notified on new types of frauds detected and how to help preventing them. The terminal application should also always prefers chip over magnetic stripe up to the point where magnetic stripe transaction is only allowed as a fallback after chip is found to be inoperable. So on a well implemented terminal you have to try chip first and only if it doesn't work the terminal should allow you to try with magstripe, if the magstripe says that the card is chip-capable. Additionally, Visa and Mastercard both monitor rates of fallbacks (situations where magstripe is used with chip-capable cards) and issues penalties to entities that cross specific thresholds.

tialaramex 6y ago
> At every step Visa and MasterCard promotes highest possible level of security. No doubt you've drunk the Kool-Aid but this isn't true. EMV rests heavily upon cryptographic protocols, we know the highest possible level of security is achieved for such protocols by publishing them for evaluation and improvement by independent researchers. EMV didn't do that. There are many but here's a particularly low hanging piece of fruit that results from not having independent researchers looking over what you've done: It is critical to the EMV design that the payment terminal picks unpredictable numbers. These aren't just nonces, where it only matters that they're different, they must be random or a relatively cheap (worth doing to forge credit card transactions) attack succeeds. So what do you suppose those standards documents require? You might even have read it. The certification document says for testing read two of these "unpredictable" values. Ensure they are different. Guess what lots of real hardware actually deployed in the wild does to achieve that requirement? It uses a counter. Because random numbers might sometimes not be different leading to a tiny chance of failing the mandatory test, but a counter will reliably pass the test even though it's insecure. Genius. So you go from "Hypothetical adversary needs to guess a large random number" to "It's just a counter" and attack cost falls from billions of dollars to a few cents per transaction on top of R&D.

tialaramex 6y ago

> At every step Visa and MasterCard promotes highest possible level of security. No doubt you've drunk the Kool-Aid but this isn't true. EMV rests heavily upon cryptographic protocols, we know the highest possible level of security is achieved for such protocols by publishing them for evaluation and improvement by independent researchers. EMV didn't do that. There are many but here's a particularly low hanging piece of fruit that results from not having independent researchers looking over what you've done: It is critical to the EMV design that the payment terminal picks unpredictable numbers. These aren't just nonces, where it only matters that they're different, they must be random or a relatively cheap (worth doing to forge credit card transactions) attack succeeds. So what do you suppose those standards documents require? You might even have read it. The certification document says for testing read two of these "unpredictable" values. Ensure they are different. Guess what lots of real hardware actually deployed in the wild does to achieve that requirement? It uses a counter. Because random numbers might sometimes not be different leading to a tiny chance of failing the mandatory test, but a counter will reliably pass the test even though it's insecure. Genius. So you go from "Hypothetical adversary needs to guess a large random number" to "It's just a counter" and attack cost falls from billions of dollars to a few cents per transaction on top of R&D.