There are numerous articles floating around the ’Net deploring the fact that the notoriously unpleasant Andrew Auernheimer (aka “weev”) has been locked up for stealing the e-mail addresses of some 114,000 different AT&T iPad 3G customers.
The most recent one to grab my attention was a piece by Errata Security’s Robert Graham which appears to intentionally misinterpret various aspects of the U.S. Government’s case against weev in order to support the notion that the CFAA (the U.S. equivalent of the U.K.’s Computer Misuse Act) could be used to arbitrarily prosecute anybody, and makes various arguments as to why the prosecutors’ case amounts to a “liberal reinterpretation of the rules of the Internet” to find weev in violation.
Unfortunately, as with most such claims I have seen, Graham has confused the technical and legal arguments in order to arrive at the conclusion he wants to reach.
First, he discusses “User-agent rules”, implying that because the User-Agent header is not intended to be used as a means of identification, the fact that Auernheimer’s accomplice, Spitler, spoofed the iPad’s User-Agent string should be disregarded. What he fails to mention is that while the User-Agent is not supposed to be being used as a means of identification, AT&T was in fact using it that way, and both of them knew this to be the case, since they knew that the AT&T server would not respond unless the User-Agent string was set to something matching that provided by an iPad.
Second, he talks about the URL, and in particular the fact that the URL is visible to end-users in ordinary web browsers and that it is also generally modifiable by end-users. Again, the facts surrounding this particular case are omitted in favour of generalisations.
In particular, Graham’s argument fails to note that the mere existence of a web server on a particular machine most certainly does not imply that anyone anywhere is authorised to access any URL at that address. Even if some URLs at a particular address are public, there may well be private URLs that are unintentionally accessible; that does not mean that you are authorised to use them.
Moreover, the URL at issue was not public — it formed part of the carrier settings for AT&T supplied to iPad users in encrypted form and was never presented in the clear to those end users for their use. In order to obtain it in the first place, Spitler had to decrypt the iOS binary (which he was not in any case entitled to be in possession of, as he did not have an iPad), and then had to hunt through the resulting data for likely URLs.
While there is a legitimate worry that CFAA or the Computer Misuse Act could
end up being applied to a normal Internet user who had merely replaced
?articleID=13, that is very much
not what happened in this case. Rather, the secret URL in question made
use of the device’s ICC-ID to allow AT&T to pre-fill the user’s e-mail address
in a log-in form. The ICC-ID is a secret number shared between the device and
the network operator; it is a bit like a credit card number, both in length
and format, and in that the number is very likely visible on the SIM card
and/or packaging. It is also like a credit card number in that it is possible
to generate valid numbers — and in this particular instance, Spitler and
Auernheimer did so based on the knowledge that the ICC-IDs issued to AT&T
iPads were sequential in nature.
Note again — the mere fact that it is possible to generate valid numbers that are not yours does not entitle you to use those numbers to authenticate yourself, either to a bank in the case of credit card numbers, or in this instance to AT&T’s network. Nor does the fact that both credit card numbers and ICC-IDs provide relatively poor security guarantees mean that you are entitled to take advantage of that fact.
The prosecution has not, as Graham claims, insisted that mere editing of a URL is illegal. The unlawful act is the unauthorised access, and there is no question that both Spitler and Auernheimer knew that they were not authorised to harvest e-mail addresses from AT&T’s servers in this manner (indeed, as the prosecution points out, Auernheimer referred to it as “the theft” in an e-mail to a journalist).
Graham then goes on to worry that “legitimate” security researchers may be in fact no different to Auernheimer, which is a bit of a stretch. White hat security researchers, who ask permission (or are contracted by the target) before attempting any kind of exploit, could not find themselves in the mess Auernheimer finds himself in. Even grey hats, who may not ask permission first, would at least have told their target about the security hole and given them the opportunity to repair it. Auernheimer did not do that. Instead, he attempted to use the security breach for his own personal gain (in this case to promote himself and his security company), and purposely did not inform AT&T prior to handing the e-mail addresses over to the press.
This is not, as some are suggesting, a case of the U.S. Government making an “arbitrary decision that you’ve violated the CFAA”. That conclusion is simply not supported by the evidence.
Should Auernheimer be locked up? Undoubtedly, if not for this then for any number of other things he’s done in the past. What about for this specific offence? Yes, I think he should. It seems to me unarguable that AT&T did not intend for anyone to be able to garner a list of e-mail addresses from their server. It also seems unarguable that Auernheimer and Spitler knew this — i.e. they knew that their activity would constitute unauthorised misuse. Auernheimer is certainly unpleasant but he is not an idiot, and he must also have been aware that there was a risk that law-enforcement might take issue with his activities. There is simply no way to reconcile these facts with the notion that he is somehow being victimised for doing something that is in any way reasonable.