THE tiny but highly regarded e-mail host Lavabit abruptly pulled the plug on its 400,000 users on August 8th. Edward Snowden, the fugitive American intelligence contractor, was a user. Lavabit’s owner, Ladar Levison, said he would rather shut the service down than “become complicit in crimes against the American people” after receiving government instructions that he says he cannot speak about. Before posting the message, he apparently rendered his customers’ stored e-mail permanently unreachable, probably by “zeroing” disk drives (using multiple passes to prevent the retrieval of magnetic “ghosts” left behind), permanently destroying the encryption keys necessary to extract archived messages, or both. Shortly afterwards Silent Circle, a firm that offers secured audio, video, messaging and e-mail said that it had killed its own e-mail system, which relied on different technology, even though it hadn’t yet been served with legal orders by the government. How do such “secured” e-mail systems work?
The internet’s standard e-mail protocols were developed decades ago, with little thought for security. Keeping messages safe from prying eyes remains jury-rigged today. To send e-mail, a user employs software that communicates with a centralised server run by the user’s internet provider (such as ATT or Comcast), or a third party, which includes giants such as Apple, Google and Microsoft, as well as relatively tiny firms like Lavabit. Without additional configuration, the text of a message travels over the open internet from the user’s software to the server. It thence wends its way to a mailbox on the same server or via the internet to another mail server at which the recipient has his delivery address. At any point along the way, a spy or hacker with the ability to access a server’s network can read all of these unencrypted messages. Should the snooper break into a mail server, all stored and transmitted e-mail would also be ripe for inspection.
Secured e-mail makes the spy’s job harder. Over the past few years mail hosts have made it simple to switch on secure transport between a sender or recipient and a mail server. Just as with secure web-sites, an e-mail server provides a user’s e-mail software with its encryption details, which are validated by third parties (certificate authorities). These third parties are, in turn, validated by components that come built in to an operating system or software program. After validity is ensured, a one-time-use session key is created and exchanged between the e-mail program and the server, and the subsequent communications are scrambled using that key. Intercepting that stream of data is to no avail. But even these security methods are vulnerable because of the central role of the e-mail servers and the firms that operate them. Even with encrypted “tunnels” between an e-mail program and a server, the e-mail is decrypted when it reaches the server system before it is stored or sent on. Hackers can hoover up messages at that stage. Security certificates can be forged and used with other security loopholes to redirect e-mail. Services that encrypt stored messages can also be prised open. If the firm uses its own keys to encrypt the messages, or has access to encryption keys that it stores on behalf of users, it may be compelled to hand them over to a government authority, or have them stolen. As Silent Circle admits, “If we are managing the keys, we can hand over the messages.” For Lavabit, which locks away a user’s mailbox encryption key so that it may be accessed only by someone possessing the user’s account password, the password may be intercepted at the website when the user logs in to use webmail.
The recent problems faced by firms such as Lavabit have reawakened interest in a relatively old-fashioned way of keeping messages secret. End-to-end encryption, in which senders encrypt messages on their own computers using a standard like PGP (or its open-source equivalent, GPG), has been around for more than two decades. The sender must first know the recipient’s public key, a safely distributable component of an asymmetrical encryption system. The public key allows a message to be encrypted, but only the recipient may decipher its contents, using a locally stored private key. These systems are cumbersome in practice, because of the requirement to know each recipient’s public key, and so are little used outside large organisations. Even these systems leak “metadata” about the sender, recipients, message timing, origin and destinations. And a keylogger installed on a user’s computer can capture the passphrase used to lock away the private key. Nonetheless, such systems may get a second look now that the most-admired methods of securing e-mail have been exposed by their operators as offering scant protection when the government comes knocking.
Powered by Facebook Comments