Experts cast runes on Google phone security
Paranoid Android
Cloud storage: Lower cost and increase uptime
Google's plans in new mobile phone platforms may not be much further advanced than slideware, but security experts are already picking apart the initiative to look for potential holes. Interest is focused on whether or not Android will be totally open or adopt an (arguably more secure) system of signing approved applications.
Some mobile security specialists are comparing the track Google has taken with its Android platform for mobile computing with the locked-down approach adopted by Apple.
Android is still in the preliminary stages of development but Google's announcement of the platform, while thin on details, gives a basic idea of how it might shape up. Android consists of a software stack for mobile featuring an operating system (based on Linux), middleware, and some default applications like a browser. The search engine giant is promoting the platform through the Open Handset Alliance.
Google is seeking to encourage developers to write for the platform. Applications will be developed using Java and use a framework provided by Google including its own Dalvik virtual machine.
The framework is based on an open source model. Google sees this as a way to bring more openness to the traditional closed mobile environment, allowing anyone to write applications that make use of the functionality available on handsets. That means, at least theoretically, that a potentially malign application could call upon any of the phone's core functionality such as making calls or sending text messages.
The security debate has focused around the open design philosophy behind Android.
Mikko Hyppönen, chief research officer at F-Secure, writes: "The key issue here is whether Android will go for totally open systems or whether they will adopt a system for signing approved applications (such as Symbian)."
"If unsigned and unknown applications written by anyone have full access to phone features, we smell trouble."
Hyppönen is quick to emphasise that his comments are preliminary. The full specifications of Android phones will be unclear until devices based on the technology become available in late 2008. Mobile malware attacks are unlikely to follow soon after, however insecure Android-based phone may or may not be.
"It's pretty guaranteed that no criminal attacks will take place until the installed base for Android has become large enough to interest the bad guys financially. This might never happen," Hyppönen said, adding that Apple's iPhone has already achieved a significant-enough user base for iPhone malware to become a possibility.
Eric Chien, a security consultant at Symantec, who (like Hyppönen) has taken a long-term interest in mobile security, notes that Apple's decision to lock down the iPhone mitigates against the possibility of malware being allowed to run on the device (at least until it is unlocked). By contrast Android's security model is based on querying users about whether or not applications are permitted to run.
"Because Android is still in development, making predictions about the threat landscape to Android is a bit premature, but history has shown us that a prompting model is far from effective," he writes.
Chien outlines some hypothetical attack scenarios, based on what we know from the world of Windows-infecting pathogens.
"The vast majority of Windows malware requires user interaction, some of which is invoked by social engineering and much of it simply because the user isn't sure what option to choose. Imagine that you download a game and the game requests the ability to send SMS messages from your phone in order to post your high scores to a central server. You agree, but little do you know that the application is also sending SMSes to a high-cost pay number." ®
COMMENTS
No Signing please...
Signing really doesn't help for several reasons:
1. signing is a very expensive exercise if you support a lot of phones (and don't kid me that Android will be write once, run everywhere). So there goes the student/amateur/small developer set, which is the vital groundswell
2. signing won't guarantee that an app-generated SMS isn't premium rate, or that the data being sent isn't sensitive
3. signing means nothing to the end-user
My experience is that the application doesn't improve in quality by going through the signing process - the only 'faults' its exposes are very minor problems.
As far as I can tell, signing is just another way to make money out of developers. Maybe that's where Symantec comes in? :-)
If this is to be an 'open' OS/API, then the applications also need to be 'open'. Yes, there's a risk placed on the user, but not that different to that of a PC app.
The right thing to do is educate the user, not lock out developers.
The iPhone isn't worth exploiting
It's become clear that, despite what the experts say, it's not worth anyone's while exploiting the iPhone. Think about it - there was a security hole that allowed web pages to run arbitrary code as root, with a robust, well-documented and very public exploit that could be used to download and run a program supplied by the attacker's. All the tools and know-how to develop such a program exist. Yet, despite all this being available for quite a while before Apple released a patch, no-one actually bothered exploiting it for malicious ends.
amanfrommars is dead, long live a manfrommars
Blimey, what he/she/it has said seems to make slightly more sense than usual.
Surely its not the same person/bot or the bot has developed some AI!!
The Matrix cant be far away

IT infrastructure monitoring strategies
What you need to know about cloud backup
Agentless Backup is Not a Myth
Customer Success Testimonial: Recovery is Everything
Data control in the cloud