Schneier on Security: Domain-in-the-Middle Attacks

Bruce Schneier on the real world effectiveness of a very simple domain name based man in the middle attack.

Here is a Wired article on the same issue showing how it was used to steal 20 GB of email from a Fortune 500 company.

Thanks to Bruce Schneier for linking to this interesting article on using patterns in language to identify the author of emails.

While the technique would not allow them to identify your anonymous emails in an ocean of others, that is rarely the real world threat scenario.

In many cases there is a relative hand full of likely authors of a given email or group of emails. It is often possible to gather large samples of emails known and acknowledged to be from the likely authors. In that case this technique has a small group of targets and excellent training materials which allow for very high levels of accuracy (the authors of the paper claim 80% – 90%). That is probably enough to get a warrant to search your home and computers.

Unless you have been unusually careful, the gig is probably up by then. Remember, this might not be for criminal matters. It many cases this would come up in whistle blowing or other non-criminal situations.

A reader of this blog recently emailed me to ask:

What s/w do you recommend to keep anonymous while using Gmail, IE, Outlook, and Facebook on a laptop?

This is actually a very tricky question because the nature of all of these tools, except Internet Explorer (IE), is to be associated with a visible and discoverable account and identity in the “cloud”. I will discuss IE last and separately.

Gmail ties to your gmail and other Google accounts. Outlook ties to some existing email account at some email provider. Facebook is tied to your Facebook account and is explicitly designed for making your information public.

The profound question here is, what do we even mean by being anonymous using these services? I would argue that the best one can manage is to be pseudonymous; that is to maintain a persistent and visible pseudonym / alias which, while discoverable, is not associated with your true identity.

Fortunately Gmail and Facebook are free and typically do not require any real credentials to set up an account, and many of the free email providers work similarly. Using Anonymizer Universal (AU), and a browser with no history or cache to set up the accounts would ensure they were not connected to your real identity. It is important that the accounts never be accessed in any way except through AU, or they will be forever after associated with your real IP address. Furthermore, it is critical that the browser used is never used for any activity connected to your real identity, or the cookies and other digital detritus in your browser may allow these sites (or other folks) to tie the pseudonym to your other real name accounts.

IE is in many ways the easiest because there is no underlying account, but all the same rules apply. You need to ensure that you isolate your anonymous or pseudonymous activity from your real name activity.

For all of this activity a virtual machine can be a very effective tool. For example, if you use a Mac you can use a virtual machine running Windows or Linux for all of your alias activities and use the normal operating system for your real name activities. Similar tools exist for other operating systems.

RIM averts BlackBerry ban in UAE | Security – CNET News

The announcement provides very little information about what RIM did to avert the ban, whether they made significant changed (compromises) to their system, or whether the UAE blinked and backed down from the threatened ban.

This NYTimes article discusses a bill which the Obama administration is proposing to submit to congress. The general background of the bill is that evolving technology has made it more difficult for law enforcement to conduct effective wiretaps and other intercepts because much of the targeted communication now takes place on the Internet and is often encrypted.

The actual text of the proposed bill does not appear to be available, but the article lists the following likely requirements.

1. Communications services that encrypt messages must have a way to unscramble them.
2. Foreign-based providers that do business inside the United States must install a domestic office capable of performing intercepts.
3. Developers of software that enables peer-to-peer communication must redesign their service to allow interception.

The first of these is similar to the CALEA law which requires telecommunications carriers to design their services to enable automated real time intercepts. While this generally sounds reasonable when “we” say it, the idea is more ominous when coming from some other governments.

The second of these feels uncomfortably familiar. See my past blog posts (and here)on the attempts of privacy hostile countries to require similar concessions from RIM.

The third proposal is completely outrageous. In effect it says that I may not speak in a way which is unintelligible to the wire tappers. As a colleague quipped “I am hiring Navajo code talkers.” This would require a back door be inserted in to cryptography tools. Experience shows that any crypto system with such a back door will be breached and then left vulnerable to the enormous number of criminal hackers on the Internet today.

In 1993 the US Government proposed a system called the “Clipper Chip” which would provide all encryption for personal computers, but to which the US Government would have back door access. This was a terrible idea then, it was widely ridiculed, and suffered a well justified death by 1996. This third proposal would be much worse. It is asking huge numbers of non-crypto experts to build back doors in to their systems. Frankly, the cryptography in most software is already badly broken in many cases. Something as subtle and complex as a secure and effective law enforcement back door would be far beyond their abilities and render currently poor security completely untrustworthy.

All this is not to mention the potential abuse by oppressive regimes, who will pounce on the capability to further crush dissent within their countries. Finally, it will be largely ineffective against serious threats. Very strong and easy to use cryptography is already available world wide, for free (GPG, ZPhone, TrueCrypt, etc.). This is a classic case of damaging the innocent while leaving the guilty and dangerous unaffected.

It seems to me that there is a pendulum swing to these things. Technology cuts both ways. Some times it favors the interceptor and some times it favors the communicator. In most ways the Internet has been a fantastic boon to law enforcement. Cloud computing, email hosts, social networking, open WiFi, and huge hard drive that encourage people to save everything all provide law enforcement with enormous amounts of information they could never have collected in the past.

It may not be shocking to anyone that there is no federal push to make that more difficult to access while pushing to enhance their ability to intercept encrypted communications.

All this is argument about a bill we have not seen yet. Let us hope that the furor that has swirled around it will cause it to be retraced or modified significantly before it is actually delivered to congress.