Reg probe bombshell: How we hacked mobile voicemail without a PIN

[Reefa]

Special report Voicemail inboxes on two UK mobile networks are wide open to being hacked. An investigation by The Register has found that even after Lord Leveson's press ethics inquiry, which delved into the practice of phone hacking, some telcos are not implementing even the most basic level of security.

Your humble correspondent has just listened to the private voicemail of a fellow Regjournalist's phone, accessed the voicemail inbox of a new SIM bought for testing purposes, and the inbox of someone with a SIM issued to police doing anti-terrorist work. I didn’t need to use nor guess the login PIN for any of them; I faced no challenge to authenticate myself.

There was a lot of brouhaha over some newspapers accessing people's voicemail without permission, but one of the strange things about it all is that at no stage have any fingers been pointed at the mobile phone networks for letting snoops in. And some doors are still open.

It's believed the infiltrated inboxes merely had default PINs, or passcodes that were far too easy to guess, allowing eavesdroppers to easily drop by. People were urged to change their number codes for their voicemail, but, as we shall see, that advice is useless – you simply don't need to know a PIN to listen to someone's messages.

Going down the rabbit hole

The login flaw was discovered during development work I was doing on a virtual mobile phone network that's aimed at folks who struggle with modern technology: it allows, for example, an elderly subscriber to ring up a call centre and ask to be put through to a friend or relative, rather than flick through a fiddly on-screen contacts book.

In this case, the operator makes the connection between the subscriber and the intended receiver, but the "calling line identification" (CLI) shown at the receiving end is that of the subscriber and not of the call centre. CLI is the basis of caller ID in the UK, but it's a bit of a misnomer because it can be changed as required.

I’d long suspected that miscreants were hacking voicemail by spoofing their CLIs to fool the phone system into thinking it was the handset collecting the messages – but surely that's too easy? It is trivial to set an arbitrary CLI when making a call. I had to find out if voicemail systems were vulnerable to spoofing.

I was emboldened by an email from Register reader Sebastian Arcus, who had set up some software for making voice calls over the internet (VoIP in other words) using his mobile phone number, and was surprised that he was able to collect his voicemail from his VoIP client without having to hand over an access PIN. I was further goaded in a chat in the pub with Reg man Andrew Orlowski, who bet me I couldn’t hack his voicemail. I should’ve asked for money to back that one up.

How it should work and how it falls apart

If you call your voicemail service from a handset linked to the account, you go through to your message inbox without the need to enter a PIN, presumably as a convenience. Use any other phone and you are asked for a PIN access code. If there is no PIN set, you don’t get to the voicemail. So far, so good.

The special sauce here is how does the mobile phone network know which phone you are calling from? The easy way is to look at the CLI sent when establishing a call.

Unfortunately, as our reader found out, this caller identification isn’t at all secure and can be spoofed, so we looked at Three, EE (and Orange), O2 and Vodafone.

How well do the big four networks protect your private voicemail?

We set up a VoIP handset to inject the necessary code to tell the network that our handset had the mobile number of the voicemail account we wanted to hack. We then dialled the voicemail service number to see if it would let us in. All networks have two voicemail numbers: a shortcode that you use from the mobile, and a long number when you call in from another phone or sometimes when you are abroad and the shortcode doesn’t work on the network you’ve roamed to.

We could only use the long number because we were not on the mobile network under test. The issue is: what does the voicemail system do when you dial the long number from a handset which identifies itself as being a subscriber?

We’ll get the secure ones out of the way first. We couldn’t hack either Vodafone or O2, so their systems must rely on more than simply checking the CLI sequence in a call. Vodafone handles the issue best. All calls to the long number ask you both for the number of the phone you are collecting the voicemail for and for a PIN. It ignores the caller display completely.

O2 got confused. The call wasn't placed, and we got instead a message generated by our VoIP system saying the number we were calling wasn’t available. O2 uses a system where you call your own mobile number and press star when you get the intro message to get to the voicemail menu. We suspect that calling from the mobile number on the VoIP network just confused everything, the voicemail system went round and round in circles until our VoIP timed it out.

But the results with Three and EE were shocking. I’d just bought a new pay-as-you-go SIM on Three, put credit on it, and set up the voicemail, which asks for a PIN right up front. I then switched off the mobile and called it using another phone to leave a voice message.

We programmed the VoIP system to present the Three mobile number to the network and dialled the long number for collecting voicemail. We got straight through as though we were using handset with the Three SIM.

It's enough to make your scr-EE-m

It was similar with EE. Our man Orlowski has a phone on the EE network, so we programmed his mobe number into the VoIP phone and called the voicemail long number. We got straight in. Unfortunately he didn’t have any voicemail, so we called from another phone, left a message and then called back on the VoIP phone and listened to it.

Testing with an Orange number (Orange and T-Mobile UK are part of EE) was more interesting. For this, we turned to a contact who works closely with the kinds of people who legally carry concealed guns – anti-terrorist, organised crime, under-cover, witness protection and the like. They use a mix of SIMs from Vodafone and Orange. So we called him, explained what we were going to do first, and then spoofed his Orange number. We got in, but didn’t want to listen to his voicemail.

So we changed the greeting message.

Anyone calling him would learn that he had changed his name to “Mabel”. It makesfraping look tame.

We did however find that with Orange it would sometimes ask for a PIN and sometimes not. We put this down to routing. We suspect – and such things are the day-job for one of the people helping me – that when the call went through some routes its illegitimacy was spotted, and when it went through others the call went straight through as though it was kosher.

The urgent issue is for EE and Three to make their systems secure. We’ve deliberately not given blow by blow details of how to spoof the CLI, but we can’t be the only people to have figured this out.

It's not like the networks have not been warned. The majority of the information presented to the Leveson inquiry on how to hack voicemail was redacted, but in a brief public document [PDF] the danger of CLI spoofing is mentioned.

And the mobile networks' own industry body, the GSMA, also warned of the danger in its voicemail security guidelines published in February 2012. In this document the GSMA talks about fraud as well as security. It points to the danger that a crook could register a premium-rate number and then use that number to leave a message on the mark's voicemail. By spoofing the CLI, the miscreants can then pick up the message and return the call, raking in the profits from the premium-rate call.

What Three and EE must do next

There is a lot that the two networks could do. Using CLI, or at least CLI alone, is shoddy. As a telco, they get all the necessary signalling information to know if the call is coming from their network or another one. This is true even if the handset is roaming, not least so that they can charge you for the call. Networks are never shy of charging for calls. They can also look at the Home Location Register (HLR) and see if the phone calling them is actually in a call.

By using these techniques they don’t have to resort to the Vodafone system of always asking you for your number and a PIN when you call the long voicemail collection number, but they could be sure that you are who you say you are. The network also gets the cell tower ID and IMEI of the incoming call. Now these are different systems, but linking the two together would be belt and braces.

We approached Three about this, and a spokesman said: "The advice we've always given customers about security is to mandate their PIN. This is particularly so for people who worry that if a phone is stolen, it might be used to access their voicemail. This advice is given under the voicemail security pages of the Three website."

Meanwhile, EE wanted to reassure its customers that it is investigating and systems are being updated to mitigate this technical issue. EE also gave us this statement"

First and foremost it’s illegal to access a voicemail account without the owner’s permission. If any customer has concerns about voicemail security we would advise them to follow a

few simple steps

on their device and set up PIN entry.

Comment

The mobile phone networks are more than missing a trick. While they complain about how the over-the-top players, such as WhatsApp and Skype, are stealing their lunch money, they do have one thing no one else can offer: complete control over the signalling and voice path. They could offer security at a level that would command a significant premium and yet they leave the door keys under the flower pot.

Source