Updated Here we have yet another example of an internet-facing home security camera with chocolate-padlock-grade security.
The surveillance cam, examined by security firm Bitdefender, comes with motion and sound detectors, two-way audio, built-in lullabies to send children to sleep, temperature and humidity sensors and a microSD/SDHC card slot. You can stream video from it in real-time across the web, and it's supposed to be used as a baby monitor, remote-controllable home CCTV, and so on.
Its firmware does virtually nothing to protect it from miscreants around the world, we're told. When you switch it on, it creates its own unsecured Wi-Fi network so a management app running on a nearby smartphone can connect to it. Then the app tells the camera how to connect to the home's wireless network so it can reach the internet.
The home network's credentials are sent over the air from the app to the camera in plaintext, so anyone nearby snooping on the gadget's hotspot can get hold of the password to the home's private Wi-Fi.
Next, when the app connects to the device directly over the internet – such as when the owner is out at work – the software uses basic HTTP authentication to log into the gadget, essentially exposing the plaintext username and password needed to access the device.
The gizmo has a default username and password combination, although it can be changed by the owner. Either way, it can be slurped by eavesdroppers or looked up from a manual, and used by anyone in the world to connect in and spy on victims. Connections are allowed into the device from the outside world via UPnP. The firmware and app also use Base64 to encode the traffic between themselves, which is trivial to decode.
When the camera wishes to send an alert to the phone app, it contacts its backend servers using SSL and provides its hardware MAC address for authentication. However, the authentication checks are completely flawed. This means anyone can ping the manufacturer's servers over HTTPS and provide the MAC address of a stranger's device to masquerade as that gizmo.
You can potentially combine these security shortcomings to trigger a bogus alert to the phone app and capture the device's username and password login when the app tries to connect to the camera to see what the problem is, as Bitdefender explains:
Every time it starts and at regular intervals, the device sends an UDP message to the authentication server, containing device data, an ID number represented by the MAC address and a 36-character code. However, the cloud server does not verify the code, it trusts the device’s MAC address to perform the authentication.
Consequently, an attacker can register a different device, with the same MAC address, to impersonate the genuine one. The server will communicate with the device that registered last, even if it’s rogue. So will the mobile app. This way, attackers can capture the webcam’s new password, if the user changes the default one.
To speed up the process and grab the password faster, an attacker can take advantage of the camera’s push notification feature. Users can opt to receive notifications on their smartphone, specifically video alerts, whenever the camera detects any suspicious sound or movement in their homes. When the user opens the app to view the alert, the app will authenticate on the device using Basic Access Authentication and, thus, send the new password unencrypted to the hacker-controlled webcam.
Finally, attackers can enter the username, password and ID to get full control of the user’s webcam, through the mobile app.
Alexandru Balan, chief security researcher at Bitdefender, said that by changing the last six digits of the MAC address it is possible to brute-force access to other cameras, with a 9 or 10 per cent success rate, from anywhere in the world.
"In a dark way, that's the fun of it," he said. "Most IoT attacks are proximity based – you have to be in range of the device itself. But here you can hijack the camera and view its stream even through a firewall and private IP address."
George Cabau, Bitdefender's antimalware researcher, explained: "Anyone can use the app [to access a camera] just as the user would. This means turning on audio, mic and speakers to communicate with children while parents aren't around or having undisturbed access to real-time footage from your kids' bedroom. Clearly, this is an extremely invasive device, and its compromise leads to scary consequences."
Bitdefender is keeping quiet on the manufacturer's name until the issue is patched, but Balan said it was a well-known manufacturer with plenty of devices in circulation. The vendor is working on a fix now. ®
Updated to add
We understand the camera maker is Edimax – and the smartphone app is the EdiView II.