As you will probably know my Birthday post for Android Malware has deserved a mention from Engadget and Wired. Easily predictable but not for me, the Engadget link has been flooded by comments posted by Android supporters and adversaries, with possible trolls’ infiltrations, up to the point that the editorial staff has decided to disable comments from the article. The effect has been so surprising that someone has also insinuated, among other things, that I have been paid to talk s**t on the Android.
Now let me get some rest from this August Italian Sun and let me try to explain why I decided to celebrate this strange malware birthday for the Android.
First of all I want to make a thing clear: I currently do own an Android Device, and convinced, where possible, all my relatives and friends to jump on the Android. Moreover I do consider the Google platform an inseparable companion for my professional and personal life.
So what’s wrong? If you scroll the malware list you may easily notice that the malware always require an explicit consent from the user, so at first glance the real risk is the extreme trust that users put in their mobile devices which are not considered “simple” phones (even if smart), but real extensions of their personal and professional life.
You might say that this happens also for traditional devices (such as laptops), but in case of mobile devices there is a huge social and cultural difference: users are not aware to bring on their pocket dual (very soon four) cores mini-PCs and are not used to apply the same attention deserved for their old world traditional devices. Their small display size also make these devices particularly vulnerable to phishing (consider for instance the malware Android.GGTracker).
If we focus on technology instead of culture (not limiting the landscape to mobile) it easy to verify that the activity of developing malware (which nowadays is essentially a cybercrime activity) is a trade off between different factors affecting the potential target which include, at least its level of diffusion and its value for the attacker (in a mobile scenario the value corresponds to the value of the information stored on the device). The intrinsic security model of the target is, at least in my opinion, a secondary factor since the effort to overtake it, is simply commensurate with the value of the potential plunder.
What does this mean in simple words? It means that Android devices are growing exponentially in terms of market shares and are increasingly being used also for business. As a consequence there is a greater audience for the attackers, a greater value for the information stored (belonging to the owner’s personal and professional sphere) and consequently the sum of these factors is inevitably attracting Cybercrooks towards this platform.
Have a look to the chart drawing Google OS Market share in the U.S. (ComScore Data) compared with the number of malware samples in this last year (Data pertaining Market Share for June and July are currently not available):
So far the impact of the threats is low, but what makes the Google Platform so prone to malware? For sure not vulnerabilities: everything with a line of code is vulnerable, and, at least for the moment, a recent study from Symantec has found only 18 vulnerabilities for Google OS against 300 found for iOS (please do no question on the different age of the two OSes I only want to show that vulnerabilities are common and in this context Android is comparable with its main competitor).
Going back to the initial question there are at least three factors which make Android different:
- The application permission model relies too heavily on the user,
- The security policy for the market has proven to be weak,
- The platform too easily allows to install applications from untrusted sources with the sideloading feature.
As far as the first point is concerned: some commenters correctly noticed that apps do not install themselves on their own, but need, at least for the first installation, the explicit user consent. Well I wonder: how many “casual users” in your opinion regularly check permissions during application installation? And, even worse, as far as business users are concerned, the likely targets of cybercrime who consider the device as a mere work tool: do you really think that business users check app permission during installation? Of course a serious organization should avoid the associated risks with a firm device management policy before considering a wide deployment of similar devices, most of all among CxOs; but unfortunately we live in an imperfect world and too much often fashion and trends are faster (and stronger) than Security Policies and also make the device to be used principally for other things than its business primary role, hugely increasing risks.
This point is a serious security concern, as a matter of fact many security vendors (in my opinion the security industry is in delay in this context) offer Device Management Solution aimed to complete the native Application Access Control model. Besides it is not a coincidence that some rumors claim that Google is going to modify (enhance) the app permission security process.
As far as the second point is concerned (Android Market security policy), after the DroidDream affair, (and the following fake security update), it is clear that the Android Market Publishing (and Security) model needs to be modified, making it more similar to the App Store. There are several proposals in this context, of course in this place is not my intention to question on them but only to stress that the issue is real.
Last but not least Sideloading is something that makes Android very different from other platforms (read Apple), Apple devices do not allow to install untrusted apps unless you do not Jailbreak the devices. Android simply needs the user to flag an option (By The Way many vendors are opening their Android devices to root or alternate ROMs, consider for instance LG which in Italy does not invalidate the Warranty for rooted devices) or HTC which, on May 27, stated they will no longer have been locking the bootloaders on their devices.
So definitively the three above factors (together with the growing market shares) make Android more appealing for malware developers and this is not due to an intrinsic weakness of the platform rather than a security platform model which is mainly driven by the user and not locked by Manufacturer as it happens in case of Cupertino.
Not even a week after the light version of DroidDream, a new nightmare rises from the Android Market to menace the dreams of glory of the Google Mobile OS (which has just confirmed his #1 Rank on the comScore April 2011 U.S. Mobile Subscriber Market Share Report).
Curiously, also the new malware, discovered by F-Secure, and dubbed Android/DroidKungFu.A, “has its roots” on DroidDream since it uses the same exploit, rageagainstthecage, to gain root privilege and install the main malware component.
Once installed, the malware has backdoor capabilities and is able to: execute command to delete a supplied file, execute a command to open a supplied homepage, download and install a supplied APK, open a supplied URL, run or start a supplied application package.
Of course, who is familiar with Android malware may easily imagine the next step of the infection: the malware is in fact capable to obtain some information concerning the device and send them to a remote server: The collected information include: IMEI number, Build version release, SDK version, users’ mobile number, Phone model, Network Operator, Type of Net Connectivity, SD card available memory, Phone available memory.
In few words, the device is turned into a member of a botnet (without realizing it we are closer and closer to Phase 4 of Mobile Malware, consult slide 9 of my presentation for the different phases of Mobile Malware).
Guess where the malware was detected first? Of course from some parallel Markets in China, at least according to some Researchers of the North Carolina University who detected two infected applications in more than eight third-party Android app stores and forums based in China. Nothing new under this sun of June. Luckily the researchers haven’t found infected apps in non-Chinese app stores… At least so far.
As previously stated DroidKungFu takes advantages of the same vulnerabilities than DroidDream, but this time the situation seems to be much worse. As a matter of fact it looks like DroidKungFu is capable of avoiding detection by security software.
The malware makes its best with Android 2.2 and earlier, but the owners of later versions of Android are not entirely safe: the security patches severely limit DroidKungFu, but the malware is still able to collect some user data and send them to a remote site.
Again, follow basic, common-sense guidelines for smartphone security in order to mitigate the risks of infection (here you may find some useful suggestions), even because Google Wallet is at the gates and I dare not even think to the aftermaths of a malware leveraging vulnerabilities on the Secure Element…
- DroidDream is Back! (paulsparrows.wordpress.com)
One of the most surprising things I noticed concerning the Lockheed Martin Affair, was the affirmation contained in the Reuters Article, made by Rick Moy, president of NSS Labs, indicating that the initial RSA attack was followed by malware and phishing campaigns seeking specific data to link tokens to end-users (an indirect evidence of the same authors behind the infamous RSA breach and the Lockheed Martin attack.
My initial surprise only lasted few seconds, since, this year is showing us a brand new role for the phishing attacks which are more and more targeted to steal corporate sensitive data, and constitute the first level of attack for Advanced Persistent Threats.
At first sight could be quite difficult to believe that users are still tricked by old-school phishing techniques, but a deeper analysis could show in my opinion, a possible (in part psychological) explanation relying on the fact that the users themselves are still used to think to phishing as something targeted to steal personal information (often with pages crafted with gross errors), and seems to be unprepared to face the new shape of phishing which targets corporate information with cybercrime purposes and industrial methods, which definitively means to perpetrate the attack with plausible and convincing methods, and most of all leveraging arguments the user hardly doubts about (I could doubt of an E-mail from my bank asking me to provide my account and credit card number, maybe, most of all in case I am not an infosec professional, I could feel more comfortable in providing my username to a (fake) provisioning portal of my Company).
But my information security beliefs are falling one after the other, and after reading this really interesting article by Adrienne Porter Felt and David Wagner of the University of California (the marvelous LaTeX layout!) I can only confirm that mobile devices will be next frontier of phishing.
According to this paper the risk of a success of a phishing attack on mobile devices is dramatically greater than traditional devices due to some intrinsic factors such as the smaller size of the screen, the fact that many applications embed or redirect to web pages (and vice versa some or web pages redirect to applications), the fact that mobile browsers hide the address bar, and most of all the absence of application identity indicators (read the article and discover how easily a fake native application can resemble completely a browser page) which makes very difficult to discover if a certain operation is calling a fake application on the device or it is redirecting the user to a fake application resembling a legitimate login form.
Moreover, the intrinsic factors are worsened by (as usual) the user’s behavior: as a matter of fact (but this is not a peculiarity of mobile devices), users often ignore security indicators, do not check application permissions and are more and more used to legitimate applications continuously asking for passwords with embedded login forms and. Last but not least I would add the fact that they are not still used to think to mobile applications as targets of phishing (Zitmo Docet).
Guess what are the ideal candidates for Mobile Phishing attacks? Easy to say! Facebook and Twitter since they are the most common linked applications used by developers to share their creations (the power of free viral marketing!).
Given the speed with which these devices are spreading in the enterprise (see for instance this GigaOM infographic), there is much to worry about in the near future. An interesting solution could be the operating system to support a trusted password entry mechanism. Will SpoofKiller-like trusted login mechanisms be our salvation as the authors of the paper hope?
- More Random Thoughts on the RSA Breach (paulsparrows.wordpress.com)
- Mobile Phones Are Great for Phishers, Researchers Find (pcworld.com)
The news of the day is undoubtedly the discovery that Apple devices are a bit ‘too nosy’ and regularly record the position of the device into a hidden (!!) unencrypted and unprotected file.
The unwelcome and serendipitous discovery, which was announced today at Where 2.0, has been performed by two researchers, Alasdair Allan and Pete Warden, while they were working on a project concerning visualization of Mobile Data. It looks like this unrequested feature has been introduced since the arrival of iOS 4.0 and allows the locations and their relative time stamps to be written on an easily accessible file on the device and, even worse, backed up on every PC the device has been synchronized with.
Even if the purpose of the file is unknown (at least so far), and would be appropriate to wait a reply from Apple (if any) before coming to any conclusion, this event, once again, brings to the fore privacy issues for mobile devices, strictly related to the security model for these devices, and, more in general, to the cultural approach and revolution users must face (and get used to) when dealing with mobile technologies.
For sure the main issue here is the lack of respect by Cupertino towards the users (customers?). We know that this is not the first time that a mobile applications attracts criticism for the use of private data (think for instance to the affair of Google Latitude). In the case of Apple Equipment (differently from the creature of Google) the user may not explicitly approve the sharing (would be better to say the tracking since there is no evidence of sharing so far) of his data. But even if we do not consider the ethic point of view, from a security perspective the event has a devastating impact: if the file containing the data may be easily accessed, this means that, in case of theft, could be quite easy, for a malicious user, to grab the data and reconstruct the habits of the users. If we think, for instance, to industrial espionage, this occurence has a dramatic consequence enhanced by the evidence that this kind of devices are often used by CxOs. (Who are the most targeted by the risks of consumerization of IT, of which this is yet another example).
Moreover, in most circumstances I discussed the risks of geolocation (and its correlation with users’ habits) and the importance that this data could have if massively stolen (for instance by mean of a Mobile Botnet) by Cybercrooks and conveyed to a C&C Server. In a similar scenario bad guys capable of stealing such a similar amount of data would have no difficulty at all to organize an auction “to the death” between hungry marketing agencies, which would pay gold to put their hand on them. I must admit that the thought that these “bad boys” could be just the manufacturers of my iPhone (luckily I own an Android) does not make me feel very comfortable. This situation is also paradoxical: many security vendors offer privacy advisors for (other) mobile platforms, but the evidence that one user should defend his privacy from the manufacturer itself sounds absurd and frustrating. Of course I continue to repeat that it is better to wait for an Apple official reply, but, honestly speaking the fact that these data are only available for devices provided with a cellular plan, sounds very strange.
Meanwhile, if you want to know more and enjoy (I hope so) to verify where have you been since you bought your brand new iToy, you may have a really interesting look at this link where the authors of the discovery posted an app to unleash the file and graphically map the positions.
Last but not least, there is no evidece (so far), of a similar “Feature” on the Droids.
On the other hand, these are tough times for the privacy of smartphones owners. As a matter of fact, quite curiously, today another, apparently unrelated, piece of news coming from the opposite site of the Ocean caught my attention. It concerns Michigan State Police, which has been using data extraction tools to collect information from the cell phones of motorists detained for minor traffic infractions. This has been possible by mean of Cellebrite, a mobile Forensics Tool capable to perform:
“Complete extraction of existing, hidden, and deleted phone data, including call history, text messages, contacts, images, and geotags. The Physical Analyzer allows visualization of both existing and deleted locations on Google Earth. In addition, location information from GPS devices and image geotags can be mapped on Google Maps,”
Even if the latter issue raises the question concerning to what extent the law can go when facing privacy of the citizens, the two news have in common the (mis)use of mobile data and I could not help but thinking that mobile data are continuously under attack and users should consequently consider carefully the usage of their devices (this is the reason why I used the term of cultural revolution).
Who knows, maybe Michigan State Police hoped to make further fines for speeding after detaining the motorists by tracking GPS position and timestamps. Probably if they had known the existence of the above mentioned feature of iOS, they would have avoided to buy the software and grab directly the data… At least for iOS 4 users…
Ho pubblicato su Slideshare la relazione da me redatta della Tavola Rotonda “Mobile Security: Rischi, Tecnologie, Mercato” tenutasi il 14 marzo a Milano all’interno del Security Summit 2011.
La relazione, che ho inserito all’interno di un thread del gruppo Linkedin Italian Security Professional, è visibile al link sottostante. Ancora un grazie al gruppo che ha ospitato questo interessantissimo appuntamento!