Analysis At Black Hat yesterday, security consultant Lukas Grunwald of German company DN-Systems demonstrated the cloning of a biometric passport, observing beforehand to Wired that the "whole passport design is totally brain damaged." But should we be surprised? Not exactly, because that's precisely what it says on the tin.
Grunwald boned-up on ICAO (International Civil Aviation Organisation) documentation, bought an ePassport reader and reading software, read a passport (German, but other ePassports would do the trick too), then cloned it. We should however be clear about what he has done here - he hasn't cracked anything, but he has brought the fundamental flakiness of the ePassports that are now shipping to wider attention. People will no doubt be appalled, but they could just as easily have been appalled some considerable distance back in the production process because that really is what it says on the tin.
The ICAO documentation Grunwald consulted is publicly available, and explains the detail of the various levels of security of the ePassport system, the baseline level being something not unadjacent to zero. For standard ePassports including chip and facial biometric the ICAO assumption is that an open passport can be taken as the bearer's acceptance that the passport is willingly being made available for the data to be read, ICAO's intent here being to duplicate as closely as possible the inherent Ts & Cs of traditional passport inspection systems. But the ePassport is RFID, and therefore vulnerable to skimming and eavesdropping (i.e. being read by a concealed reader and/or having the transaction between passport and 'official' reader snooped on.
Two mechanisms will be used in ePassports to impede this; first, there is the 'tinfoil hat', a mesh of metal in the cover that blocks access to the chip when the passport is closed, and second the machine-readable zone (MRZ) of the passport. The MRZ is designed to be read visually when the passport is open, and this is then compared to the copy of the MRZ held on the chip. If the two match, then the data on the chip can be read.
There are other, optional levels of security that we'll go into shortly, but what we've covered so far is what most countries will be shipping in this generation, and what Grunwald had to deal with. Here what he did again, in slow motion this time.
Grunwald bought an official inspection reader (N.B. this is legal, and even if it weren't the volumes of machines the market will need would make it trivial to obtain one) and placed his passport on top of it. Using Golden Reader Tool software from secunet Security Networks he read the chip in the passport. Golden Reader Tool is again freely available, and is widely used in the current round of ePassport interoperability testing. From there, Grunwald was able burn the data onto a chip in a blank sample passport page, giving him a blank document that looks to readers like the original passport.
Note that there's nothing particularly special about the official reader here, so it would be feasible with this level of security to use a homebrew reader. Note also that this is precisely what ICAO says you can do if this level of security is all that's used. MRZ comparison: "Adds (minor) complexity. Does not prevent an exact copy of chip AND conventional document."(PKI for MRTDs offering ICC Read-Only Access V1.1)
So what can you do with this? You've got an exact copy of the chip from one person's passport, but you do not at the moment have a mechanism for changing the data on the chip, and in order to produce an entire copy of the passport you'd need to get over the more conventional speedbumps to forgery in the rest of the document. But you do have something that's potentially quite useful, and under certain circumstances can brush aside what border security exists.