writes "Two researchers have found a way to run unauthorized code on an iPhone remotely. This is different than 'jailbreaking,' which requires physical access to the device. Normally applications have to be signed cryptographically by Apple in order to run. But Charles Miller of Independent Security Evaluators and Vincenzo Iozzo from the University of Milan found more than one instance in which Apple failed to prevent unauthorized data from executing. This means that a program can be loaded into memory as a non-executable block of data, after which the attacker can essentially flip a programmatic switch and make the data executable. The trick is significant, say Miller and Iozzo, because it provides a way to do something on a device after making use of a remote exploit. Details will be presented next month at the Black Hat Conference in Las Vegas."
The attack was developed on version 2.0 of the iPhone software, and the researchers don't know if it will work when 3.0 is released.