The FBI and Scotland Yard versus Anonymous: Security Lessons
A lot of people are fascinated by the news story that Anonymous managed to listen to a conference call between the FBI and Scotland Yard. Some of the interest is due to marvel that two such sophisticated organizations could be had, some is due to schadenfreude, and some is probably despair: if the bad guys can get at these folks, is anyone safe? To me, though, the interesting thing are the lessons we can learn about what’s wrong with security. Many of the failures that led to this incident are endemic in today’s world, and much of the advice we’re given on what to do is simply wrong or arguably even harmful.
The first issue is how Anonymous managed to record the call. The ways we’d see it done in a movie — tapping a phone line or listening to law enforcement official’s cell phone — are comparatively difficult to do. They’re not impossible, but they’re not the easy way for a task like this. Rather, what appears to have happened is what most outside security experts immediately suspected: Anonymous read an email giving the details of the call, and simply dialed in, in the same way as the intended participants. The message was sent to "more than three dozen people at the bureau, Scotland Yard, and agencies in France, Germany, Ireland, the Netherlands and Sweden;" a single security flaw anywhere along the chain could have resulted in the leak.
Here we see the first flaw: the call details were, effectively, a shared credential. It is quite probable that the conference call moderator had no idea who had dialed in. We see the same phenomenon with role accounts: many people share the password for the login, email access, etc. It may happen in the large — email@example.com — it may happen when a vacationing executive gives a secretary the password to his or her email account; it may happen when spouses or romantic partners share passwords. Whatever the reason, it creates a security risk.
Reading further into the article, we see that "One recipient, a foreign police official, evidently forwarded the notification to a private account". At that point, it’s tempting to blame that official, say he or she was poorly trained or disobedient, and stop worrying. Apart from the self-evident fact that a single security lapse shouldn’t compromise everything (a proposition easier to state than to make happen), I strongly suspect that this unnamed official was behaving very rationally: he or she either wanted email access that was too inconvenient via the proper mail servers, or wanted a different human interface. If this person had no access to work email from home, or felt that, say, gmail was enough better that their productivity was improved, it’s not surprising that this would happen. It shouldn’t happen — and one would hope that a police official working on cybercrime would understand the risks — but in a strong sense the failing was organizational: if my hypothesis is correct, they may have failed to make it easy for people to do the right thing. Let me stress this: a security mechanism that is so inconvenient that it tempts employees to evade it is worse than useless, it’s downright harmful. (Note well: I’m not saying that this official did the right thing; I’m saying that organizational policies or technologies may have led to too much temptation for people who are trying to be more productive.)
But how did Anonymous know which outside email account to monitor? This article notes that assorted groups have made a habit of targeting law enforcement email servers, with some success against less-sophisticated police organizations. That would yield a list of email addresses, and perhaps passwords. Perhaps more importantly, it can show who was using an outside mail server, one that isn’t protected by VPNs, firewalls, one-time passwords, and the like. At that point, the attackers have several ways to proceed.
First, they could try this law enforcement email password against the outside mail server. The odds are high that it will succeed; far too many people reuse passwords. And why do they do this? Because they have too many passwords to remember, especially if they’re all "strong". And of course, people are forbidden to write them down.
Most of the advice we get on security starts with "pick a strong password". (Look at CERT’s advice: the very first thing it tells people to do is "always select and use strong passwords". Patches, a really effective defensive measure, are mentioned fourth.) Strong passwords are not a bad idea, but you’re in much more trouble if you reuse passwords. No one can possibly memorize all of the passwords they have; reuse is the usual answer.
A second way in which the attackers could have compromised the official’s account is via a spear-phishing message, booby-trapped to install a keystroke logger. That’s been seen, though more often in a national security context. If the attackers did this, even encrypting the emails wouldn’t have helped; the same malware that stole the login password could probably steal the private key as well. But I’m pretty sure that no encryption was employed; most encryption systems are too hard to use. Smart-card based decryption would have helped (though such things are far less convenient to use); though there are still attacks, they’re more involved, and arguably less available to a group like Anonymous.
It’s clear that there wasn’t a single failure involved; in particular, the crucial mistake of forwarding work email to a personal account was quite plausibly a rational response to organizational policies. Preventing recurrences of this kind of incident will not be easy; there are too many weak spots.