When I started writing the stories about NXP and the security of the Mifare Classic chip, I was lucky enough to be granted an all access interview with Karsten Nohl, one of the researchers who discovered the security issues.
Q:What got you interested in RFID?
A: RFIDs fascinate me for the challenges they bring about; in particular for privacy and security. Optimally, RFIDs would be as secure as alternative ways of authentication, also resistant against new attacks specific to wireless technologies, not disclose any information about the their users, and finally cheap enough for a wide range of applications. These obviously contradicting demands and the trade-offs they require is what motivates my research.
Q: What made you want to look closer at the MiFare?
A: Our group started on Mifare to find whether its security is sufficient for the large number of applications it is used in. As a motivation for my research in which I build more secure cryptography for similar applications, I hoped to find Mifare to be weak. We had no idea, however, that it would turn out to be so fundamentally flawed.
Q: The Dutch government held hearings and compiled a report on their findings. Can you explain this and give your thoughts on the final findings by the Dutch government?
A: The discussion in the Netherlands is far from over with new hearings beginning this week, but the outcome is already astonishing: politicians are calling for proprietary technology to be replaced by open designs (and open source software!), the public transport operator has started the migration to a more secure system, and industry starts working closer with universities and "hackers" to make current and future systems more secure.
Q: NXP claims they have multiple layers of security. With that in mind, they point out that you have only one part of the algorithm. Is that true? If it is, is obtaining one part of the algorithm sufficient to gaining access to the rest?
A: I am not exactly following NXP's rhetoric in this point. They have different products, many of which are very secure. The security of the particular card we analyzed, however, relies entirely on its cryptography which we found to be weak. The recent announcement of a successor product suggest that NXP is starting to acknowledge this fact.
Systems in which the cards are used might still have other layers of security that potentially mitigate attacks. Credit cards are an example where the cards can easily be copied but fraud is often detected in the backend system. The idea of fraud detection as a security layer appears to be limited to credit cards, however, where each transaction carries a large amount of personal data and to also limited to the States where privacy laws tend to be laxer. In systems where no profiling data is available like building access control and in countries where storing such data is not permitted, fraud detection cannot provide an extra layer of security.
Q: What are your thoughts on security by obscurity? Is NXP using this method of protection?
A: Security-through-obscurity hardly ever works. The lack of proper peer-review often even hurts the security of the system. Our Mifare work discovered several vulnerabilities that could be fixed without increasing the cost of the cards. NXP did for a long time rely on obscurity for the security of some of their products, but now decided against this outdated design approach and instead bases the security of newer RFID cards on publicly scrutinized cryptography and independent evaluations.
Q: Can you explain "Kerckhoffs Principle" and why it applies to your work?
A: Kerchoff, who lived in the 19th century, observed that keeping anything secret is really hard. So instead of relying on the secrecy of your whole system, it would a lot easier to only rely on the secrecy of a small secret key. Security systems should hence be publicly known and analyzed, and only the key should be secret. When properly realised for RFID cards, Kerchoff's principle means that by analyzing their own cards, thieves cannot compromise your cards. This is contrary to our Mifare work, where we only analyzed a few copies of the the secret algorithm that is found in all cards and were consequently able affect the security of all the other billion cards out there.
Q: "This does not breach the security of the overall system," NXP is quoted as saying. "Even if one layer were to be compromised, other layers will stop the misuse." However, after reading your research, did you not prove this statement wrong? If not please correct me.
A: Cryptography can only ever be one layer in a security system, two of the others being automated fraud detection and law enforcement. Computerized systems tend to rely on the cryptography, however, and are much more vulnerable to attacks once this layer of protection is lost. Fraud detection as on often cited second line of defense appears to be limited to very specific applications where personal data about individuals is available.
Q: How likely is it that someone is at risk because of this discovery?
A: The discovery itself puts nobody at risk since we are using our insights purely academically and will only share it with operators that want to assess the security of their systems independently from our analysis. The news of these systems being insecure might, however, inspire criminals to find the secrets in ways similar to our project, but that's far outside of my area of expertise.
Q: Covering the absolute worst case, what are some other areas that could face potential harm or risk because of the flaws in early RFID technology?
A: The media attention has so far been focused on low-security application such as public transport. I see a much higher risk in access control systems to high-security areas such as military installations, chemical plants, biological agent storage, power grid control rooms, hydroelectric plants and so forth. Furthermore, secondary defenses such as fraud detection do not apply in these scenarios and systems should start upgrading shortly or at least add additional factors of authentication.
Q: In your opinion, do you think that the bottom line for business is why chips such as the MiFare are chosen over more secure but costly alternatives?
A: I think it's a combination of that and the fact that different products are hard to compare. If the manufacturer assures you that even the cheapest alternative has proven secure for more than a decade and provides "advanced security levels" wouldn't you go with it? This is yet another reason why the security of these systems has to be evaluated independently. I hope that independent security assessments and open design will become a natural requirement for security systems.
Q: The privacy issues with RFID were covered early on, now with the advent of pay cards for transit systems, do you think the financial sector will switch chips for stronger security?
A: The financial sector is very inconsistent in that respect. Bank cards in Europe have, with few exceptions, always been build around strong cryptography. Credit cards that are particularly popular in the States, on the other hand, were notoriously easy to copy and this trend seems to continue for RFID-enabled credit cards. The better cryptography that is underway for credit cards will not change this in terms of privacy but rather improve the protection against financial fraud. And even credit card companies will not be able to counter new insecurities with ever broader data harvesting ad infinitum, but rather limit the amount of personal data they keep eventually. For credit cards and other systems alike, more security and more privacy can be achieved simultaneously through the proper use of strong cryptography.
Q: Wi-Fi research led to the discovery of its flaws, and new methods of fishing out networks. (War-Driving) Can such methods apply to RFID readers? In a worse case scenario, can someone use a wide range reader to collect RFID information on a massive scale?
A: That's a constant point of discussion between industry and academics. Being a member of the latter camp, I strongly believe that large-scale attacks on the privacy of individuals will happen. Collecting private information can be very profitable and RFID tags will be the easiest way to recognize these individuals. Similar to the discussion on war-driving, industry still claims that physical constraints of the technology limit the distance from which RFID cards can be read. But similar to Wi-Fi where the current distance record is at almost 40 km, RFID equipment will improve to a point where attacks across a room become possible and even likely.
Q: Knowing what you know about MiFare, will you look into some of the other higher grade chips and see what their security looks like?
A: Many of the better chips use publicly known cryptography, so there is nothing to do for us. One system to look at are RFID-enabled credit cards, that keep all their specifications secret. It will be interesting to see when the consequences of Kerchoff's principle starts unfolding for credit cards.
Q: After reading the press about your research, do you feel the media reported this story accurately? Why or Why not.
A: For the most part, the coverage was accurate, the only exception being the connection between our work and credit cards.
Q: What would you like to clear up regarding your work, and the media coverage?
A: We haven't compromised anything about credit cards and the backward-compatible design of radio-enabled credit cards suggest that there isn't really anyhing to compromise ;).
Q: Do you have anything to add or anything to re-hash about the research?
A: As much as I criticize the security of some smart carts, I strongly believe in the potential of RFID technology to improve security in many domains. The current set-backs are hopefully only part of a learning process from which better systems will arise. Our work has so far lead to the announcement of a more secure low-priced card, the migration of several large systems to more secure technology, and the call for more open designs. These are outcomes that we couldn't even have dreamed of when we started this project.
Q: The Organization for Applied Scientific Research (known as TNO)remains convinced that the costs of building a key cracker, in dollars and time, still is far higher than the your team says. Can you explain what was used in generic terms and cost. I say generic because I do not want to disclose all of your research, nor do I think that is wise.
A: When TNO analyzed the security of the system they did not have knowledge of the secret algorithm, and while we indicated that we were always willing to share our findings for the sake of an independent investigation, they did not take us up on our offer.
The Mifare stream cipher is simple and its key is short. This alone should tell anybody that secret keys can be found cheaply. To finally end the discussion about how cheaply exactly, we made public a new attack on the cipher today that exploits its weak structure. Bottomline: The computer you are reading this e-mail on can find secret keys in at most an hour. In the latest response to this information, TNO acknowledges that our estimates are in fact correct and that systems should migrate to more secure cards as soon as possible.
Q: "TNO says the researchers' claim to have fully recovered the MiFare Classic encryption algorithm can't be fully verified. But if they haven't, TNO expects them to do so in the near future." Can you clear this up once and for all.. Do you have the whole key?
A: There is no such thing as partially discovering an encryption algorithm. Either you know the whole algorithm or you don't know anything. We only partially revealed the algorithm, however to prevent it from being known to those with malicious intent.
Q: In a March 6 statement, NXP says it "has established an open dialogue with the researchers and is evaluating possible attacks and countermeasures that could be taken in an overall systems to prevent those." What have you shared with them and how are they reacting to your work?
A: We have been talking to NXP ever since news about our research first surfaced in January. We share our insights before publicly announcing them to give NXP sufficient time to prepare responses.
It should be noted these questions were asked and answered before the release of the Mifare Plus by NXP.
See the other articles for more details.