It didn’t take long for the blogosphere to respond to research presented on Wednesday that detailed a file in Apple iPhones and iPads unknown to the vast majority of its users that stored a long list of their time-stamped locations, sometimes with alarming detail.
On Thursday, a forensics expert who sells software to law enforcement agencies gave a first-hand account why scrutiny of the location-tracking database is crucial. Alex Levinson, a forensics expert specializing in mobile devices, blogged that “geolocational artifacts were one of the single most important forensic vectors found on” the devices. As a result, he wrote a proprietary program called Lantern that law enforcement agencies use to actively examine the contents of the iPhone location database.
From The Register
It didn’t take long for the blogosphere to pooh pooh research presented on Wednesday that detailed a file in Apple iPhones and iPads unknown to the vast majority of its users that stored a long list of their time-stamped locations, sometimes with alarming detail.
On Thursday, a forensics expert who sells software to law enforcement agencies gave a first-hand account why scrutiny of the location-tracking database is crucial. We’ll get to that in a moment. But first, let’s take a sampling of the rampant naysaying.
The most common criticism was that the contents of the SQLite file, which is stored on the phone and on any computer backups, were wildly imprecise. Blogger and web developer Will Clarke, for instance, used the researchers’ freely available software to map the coordinates gathered by his own iPhone during a recent round-trip bike tour he took from Philadelphia to New Jersey. When he compared the results to the actual route, he found that “almost all the points were way off.”
In an interview with The Reg, he said some of the points on the resulting map were as much as 3,000 meters, or almost two miles, away from his true location.
“The data that is exposed basically reveals which city you were in at a given time,” he concluded in a post that called the research “sensational.” “Nothing more specific than that. It can’t tell what house you live in, it can’t tell what route you jog on, nothing like that.”
He went on to conclude: “Apple is not storing the device’s location, it’s storing the location of the towers that the device is communicating with.”
Software analyst David “Lefty” Schlesinger found similar inaccuracies when he used the database contents of his iPhone to plot a train ride he took in July from Amsterdam to Den Haag, about 60 kilometers away. He also found that the iPhone file showed he was in Santa Cruz, California, on Christmas Day and traveled as much as 80 miles, when in fact he stayed in the state’s Central Valley, some 130 miles away, the entire day.
Like several other bloggers, he also noted huge inconsistencies in the time intervals that locations were logged. Sometimes iPhones and iPads went days without updating the database, and on one occasion went almost two weeks.
The critics make a valid point that the data stored in the consolidated.db file hardly contains a historical record of a user’s real-time comings and goings, or a user’s every move, as incorrectly suggested in initial coverage from The Register and many other news sites. Researchers Pete Warden and Alasdair Allan readily acknowledge that they have yet to figure out what triggers iDevices to log location details, but it’s not unusual for hours or even weeks to occasionally pass between entries.
They said they have noted one or two grossly inaccurate locations logged in the database. One region that seems to regularly pop up in files stored on multiple phones is an area just outside of Las Vegas. Allan said the database extracted from his iPhone and the iPhones of several people he knows logged that Nevada city even though none of the owners were anywhere near it on the date indicated in the corresponding timestamp.
Las Vegas also incorrectly showed up on the iPhones of Clarke and a co-worker of his, suggesting the iOS code that logs locations may be buggy.
“We both have the exact same data point in Vegas, and neither of us have been,” he said.
Warden and Allan said their reverse engineering exercise made it impossible to learn the precise way the logging works, but they insist the conclusion of their research is still correct: The contents of the consolidated.db file stored on every iDevice and on any computer containing a backup of its data contains a “scary amount of detail on our movements.”
“By inspecting it, I can tell what part of downtown San Francisco I’m in, I can see that I’m in a particular neighborhood,” Warden said.
Added Allan: “It’s a bit above block level, but it can certainly tell that I’m in north east Manhattan, or south east Manhattan.”
They said the precise latitude and longitude plotted on a map is accurate to about 500 meters in areas where there are many cellphone nodes and as much as 4 kilometers with fewer nodes.
“It really does seem to be dependent on how good your cell coverage is,” Allan said. “If you’re in a big city like downtown San Francisco, the positioning is going to be much better. If you’re in the middle of London, the positioning is going to be much better. If you’re in a rural or semi-rural area, your positions are going to be much rougher.”
They also refuted Clarke’s assertion that the latitude and longitude coordinates logged in the database referred to the position of cell towers rather than the Apple devices themselves. Some of the extracted databases they examined plotted literally thousands of unique coordinates in a small part of a single city. It’s almost impossible that there could be that many corresponding nodes in such a confined area, they said.
What’s more, the geographic locations of cell towers is usually kept secret by the carriers who own them, and there’s no clear way an iPhone would be able to detect its longitude and latitude anyway.
“Our current stance is that this is the position of the device,” Allan said. “There has to be now or very soon a big public debate about location data and privacy. This (research) might be something that helps kick that debate off.”
Chris Soghoian, a security and privacy researcher with no connection to Warden and Allan’s work, agreed.
“I don’t think users had any idea that this information was being collected,” he said. “The fact that it doesn’t detail the exact street corner you were on and merely deals with what neighborhood you were in, I don’t think that’s going to be comforting to people.
He compared the the iPhone and iPad’s tracking of location information to the Google Street View debacle, in which roving vehicles throughout the globe logged unencrypted Wi-Fi traffic and dumped it into a giant database, contradicting previous assurances from the company. Google later pledged to destroy the data, which may include passwords and other sensitive information.
Soghoian said Apple had a responsibility to let customers know the type and extent of the information their iPhones and iPads were collecting about them.
“When you get stopped by the police and they arrest you for any crime, they can search your phone and get any data off of it,” he said. “This is definitely something that people should be concerned about and I think what it points to is that Apple isn’t taking privacy seriously.”
Indeed, Alex Levinson, a forensics expert specializing in mobile devices, blogged here that “geolocational artifacts were one of the single most important forensic vectors found on” the devices. As a result, he wrote a proprietary program called Lantern that law enforcement agencies use to actively examine the contents of the iPhone location database.
“Within 24 hours of the iPhone 4’s release, we had updated Lantern to support forensic analysis of iOS 4.0 devices,” he wrote. “Within 36 hours, we had begun writing code to investigate consolidated.db. Once a jailbreak came out for iOS 4, I wrote a small proof of concept application to harvest the contents of consolidated.db and feed it to a server for remote location tracking.”
Levinson also said iPhone location tracking has gone on much longer than indicated by Warden and Allan, who claimed it began with the introduction of Apple’s iOS 4 in late June. In fact, said Levinson, earlier iPhones contained a hidden file called h-cells.plist that contained much of the same baseband radio locations that consolidated.db has now.
“Through my work with various law enforcement agencies, we’ve used h-cells.plist on devices older than iOS 4 to harvest geolocational evidence from iOS devices,” wrote Levinson, who is a lead engineer for Katana Forensics.
Based on Levinson’s account, it’s hard to put much credence in critics who cite bugs and a lack of geographic granularity to argue that the undisclosed tracking of iPhones and iPads is harmless or inconsequential to its millions of users. Inclusion of the database means that anyone who ever loses his device risks exposing potentially large amounts of information about where he was over months or years.
That could be devastating for people embroiled in messy lawsuits or those whose whereabouts are closely guarded secrets, such as volunteers who work with victims of abusive spouses.
Of course, none of this speculation would be necessary if Apple would come clean about exactly how the location tracking it built into its devices works and what precise information is collected. The company, in keeping with its Jobsian obsession with privacy, has yet to utter a peep despite widespread media coverage.
Here’s hoping Apple’s location tracking isn’t as big a threat as some believe. But until those who know for sure speak up (Apple PR, are you listening?), we think the prudent thing to do is assume it is.