The Carte Bleue
In the early 90s France switched to a "chip and PIN" credit card system known as the "Carte Bleue". From 1993, when you used a card in a shop, instead of just swiping the card and signing the receipt, you placed the card inside a reader and entered your PIN. The payment was only authorised if the reader recognised the card as valid, and if the card confirmed that the code was correct. The system is overseen by the "GIE cartes bancaires", a consortium of around 200 French banks and financial institutions who claimed that it was unbreakable: it was impossible to create a fake credit card. Serge Humpich proved them wrong.
A penchant for security
Humpich had a degree in electrical engineering from the INSA in Lyon, but was also very interested in computing. For 12 years he worked as a programmer in areas such as back office, order taking systems etc... One of his great passions was understanding security systems. As a student and later in his spare time he devoted many hours to understanding software copy protection schemes, telephone cards, satellite tv decoders and other similar devices. The motivation wasn't financial, he was more interested in the challenge and understanding how all these systems worked.
In 1993 Serge Humpich became interested by credit card terminals used in shops. He noticed that they did not communicate with a central verification system when a payment was authorised. His goal was to create a fake credit card that would not correspond to a bank account, but that would still be accepted by the card readers in shops. Nothing would seem amiss until long after the actual transaction, when the list of daily transactions was sent back to the GIE for processing.
A system of uncertain safety
Humpich claims that from the very beginning he knew that the system was imperfect. He believed that his knowledge of both electronics and computer science would be a decisive advantage. In fact as early as 1988 when the carte bleue system was still being planned, experts claimed that the RSA key used as part of the authentication process was too short. Although at this point the 320 bit keys used were considered unbreakable they were concerned this would not stay so for long. This 320 bit key was part of the original specifications drawn up in 1983. In the early 90s cryptography experts told the banks that it was folly to believe that a system designed in 1983 would still be safe almost 10 years later, and recommended that the key length be boosted to 640 bits, which was perfectly possible from a technical point of view. It seems that economics won over security.
Reverse engineering the carte bleue
He was able to buy a payment terminal from a company that had gone bankrupt and starting analysing how it worked. After many months of patient reverse engineering Humpich managed to work out the protocol used. To get any further he needed to break the 321 bit encryption used to sign cards. By spring 1997, using a multi polynomial quadratic sieve, he had achieved this. At this point in time the GIE's yearly security budget for the carte bleue was a mere 500,000F. The hard part of the work was over, the next step was to actually produce cards using the newly discovered information. This was not difficult and does not require particularly expensive equipment. In one way the story was over, in another it had just begun.
In May 1998, after having consulted a lawyer who specialised in intellectual property issues, Humpich contacted the GIE (via his lawyer) in order to start negotiations. Humpich used a lawyer because he was keen to stay within the boundaries of the law. If money had been his main motivation it is likely he would have had little difficulty in finding some shady person willing to purchase his work. He was interested in improving the system, he wanted the GIE to admit the weaknesses of their system and improve it. Initially the GIE's reaction was one of disbelief. They dismissed Humpich as a fraud.
Proof of concept
Humpich went to the nearest train station and bought 10 booklets of 10 metro tickets with 10 of his specially crafted cards. He sent the tickets and the receipts to the GIE as proof of his success. This made them sit up, as Humpich had not only proved that fraud was possible, but from their point of view, the worst kind of fraud. Fraud due to stolen cards and the like is one thing, but this had the potential to undermine people's confidence in the carte bleue. Replacing cards and terminals with more secure ones would be a nightmare. To add insult to injury, examination of the receipts provided by Humpich revealed that the RATP, which had sold Humpich the metro tickets, had received the corresponding amount of money: the GIE had ended up footing the bill.
For several months Humpich negotiated with the GIE in order to draw up a non disclosure agreement and to fix the conditions under which Humpich would show the GIE how he had broken their system.
Unknown to Humpich, at the same time legal proceedings had been started against him. On September 17 of the same year his home was raided by the police, his equipment confiscated and Humpich was arrested. The charges against him: "contrefaçon de cartes bancaires" and "introduction frauduleuse dans un système de traitement automatisé de données" (production of counterfeit credit cards and unauthorised entry into an automatic data processing system). Initially the trial was to take place behind the closed doors; the GIE wanted to keep public awareness at a minimum.
Humpich however wanted to reveal everything to the public and on June 11, 1999 L'Est Républicain published an article describing Humpich's achievements and the legal proceedings against them. Over the next few months there was considerable pressure from the GIE to cover up all of the details. Humpich even lost his job in August. This didn't stop him giving more details to a magazine in mid September. Humpich was free in the months leading up the trial, however he was under close surveillance.
The trial started on January 21, 2000, many drew a parallel with the fact that Kevin Mitnick was released on the same day. Those who expected to find out the secrets of the carte bleue were to be disappointed: none of the technical details of the case were revealed, and there was no independent analysis of either Humpich's or the GIE's claims regarding the safety of the carte bleue system.
The position of the GIE was interesting to say the least. On the one hand they claimed that his discovery would only work in the RATP's terminal, diminishing the value of Humpich's discovery, on the other hand they wanted all of his findings and equipment destroyed. The asked for a symbolic fine of 1F, but also wanted the case to be a deterrent to others. All in all, a maze of contradictions.
Humpich's lawyers claimed that he was more of a scientist than a hacker: he could have used his discovery to make large amounts of money but he did not. He could have sold his findings to criminals, but instead he contacted the GIE. They also contested the charge of unauthorised entry into a data processing system, saying that Humpich had only worked with a terminal that was not connected to any network, hence not connected to any "data processing system".
On February 25, 2000 the court ruled that Humpich was guilty. It was decided that he knew what he was doing was wrong, that reverse engineering the algorithms used for authenticating the cards and finding the private keys qualified as unauthorised entry into a data processing system. He was fined 12,000F and given a 10 month suspended sentence. Humpich appealed, however on December 7, 2000 only a few days before the appeal was to take place he withdrew his appeal, saying that he was disgusted with the justice system and that it could not be trusted to reveal the truth. The appeal went ahead nevertheless and the original verdict was upheld. Many claimed that the GIE's only interest in the law suit was to gain access to Humpich's intellectual property without having to pay for it and to prevent further spreading of this knowledge.
In many ways this was an exercise in futility that only serve to make an example of Humpich: on February 9, 2000 a post to fr.misc.cryptologie revealed Humpich's work and methods in detail, including the private key that he had found.
Since then the key lengths used have increased, and the GIE claims that its system is safe and that only a minute fraction of transactions are fraudulent. But then again they have never said anything else.