The past few weeks a number Evil Maid attack instances have been launched against very popular drive encryption implementations. These attacks pose a very serious threat against protected data since, once launched, they are certain to succeed.
Drive encryption software, such as open-source TrueCrypt, Microsoft’s BitLocker or PGP, is used to provide the highest level of security and privacy, because all files, including any temporary files that the Operating System and applications create on the system partition, hibernation files, swap files, etc., are always permanently encrypted (even when power supply is suddenly interrupted). Such encryption involves pre-boot authentication, which means that anyone who wants to gain access and use the encrypted system, read and write files stored on the system drive, etc., will need to enter the correct password each time before the Operating System boots (starts).
According to the threat model ([1], [2], [3]), this attack exploits the same basic vulnerability as the “Cold Boot“ attack from last year, and the “Stoned Boot” attack from earlier this year, and there’s no real defense to this sort of thing. An attacker, who acquires unattended physical access to a computer system, can be confident of success in accessing the data behind any such encryption. According to the PGP blog “No security product on the market today can protect you if the underlying computer has been compromised by malware with root level administrative privileges”. Of course this is old news to the “three-letter agencies in the US Government” who last September advised Symantec’s CTO Mark Bregman to use separate laptop and mobile device when traveling to China, citing potential hardware-based compromise.
Furthermore, an encrypted drive, guarded by a user-provided password, often a simple alpharithmetic of less than 8 characters, can be easily defeated by simply guessing the secret key or stealing it through phishing, malware or domino attacks. Additionaly, the presence of an encrypted drive may be sufficient for law enforcement to demand, via subpoena, the user’s cooperation in accessing the protected data.
Facing all of the above, one could consider the concept of self-destructing data, such as the Vanish prototype, for protecting short-term information under strong encryption and plausible deniability.
