On October 16, a security researcher in Belgium named Mathy Vanhoef published some concerning findings about a vulnerability in Wi-Fi, the near-ubiquitous standard for wireless connection to the internet. The first reaction was alarmist — “The 'Secure' Wi-Fi Standard Has a Huge, Dangerous Flaw,” wrote Wired — but the second wave was more measured — “There’s a Huge WiFi Security Hole, But Don’t Panic,” wrote The Daily Beast.
The truth is, researchers have repeatedly identified serious security issues related to Wi-Fi since its inception; you are sending through the air, after all. But for various reasons, Wi-Fi has never been the criminal element’s weapon of choice. To understand why, we have to go back to the beginning.
Before there was Wi-Fi, computers were connected with cables.
The idea of a “network” — a system that would allow devices to talk to each other — was conceived and implemented differently around the world during the 20th century.
In many networks, this meant several computers would be connected to a network switch, which would physically route electrons into separate lanes of traffic to wherever they needed to go: a UCLA grad would type “LO” from a console at UCLA and be routed to a friend at Stanford, and “LOL WUT” could be sent back, with no part of the conversation ending up at the University of Utah.
These networks, the most famous of which was the ARPANET in the US that would lay the groundwork for the modern internet, built out the world’s existing telecommunications infrastructure to include physical buildings like internet exchange points, data centers, and the sprawling web of fiber-optic cables which now connect every continent that has more humans than penguins.
The Standards Committee at the IEEE, the organization of which I am a member and whose predecessor once had Thomas Edison at its helm, published the 802.11 standard we commonly refer to as “Wi-Fi” in 1999. The standard — essentially a set of instructions for devices and wireless equipment to connect with each other — allowed network equipment manufacturers to create the hardware to connect devices to a network without having to string cables everywhere.
Any hacker connected to a Wi-Fi network could potentially cause loads of mischief. Yet Wi-Fi attacks haven’t been as devastating as other types of hacking.
Within telecom infrastructure, cable is still king. But in the home and office, the physical cable’s days were numbered. Once computers became light enough to carry to the neighborhood coffee shop, people wanted to do internet things while they sipped on whatever people drank in the early 2000s. As one would expect, no one wanted to carry long cables into a public space, and few public spaces wanted to install an RJ45 jack every four feet of wall.
Wireless networks were certainly more convenient for users, but they also presented new opportunities for attackers. With a wired network, you can count on electrons traveling through circuit board traces and cables from start to finish without being copied or rerouted, (at least as long as the NSA is not involved). Doing the same way over radio waves, as Wi-Fi does, however, presents a fundamental security challenge: namely, that while in a wired network, all your secret messages, transmitted with electricity, travel inside physically sealed cables and thus, a third party can’t read them. Since radio waves broadcast over the air, it’s much easier for an observer to intercept everything being transmitted.
This has led to some problems.
The original 802.11 standard included a security protocol for authenticating connected devices and protecting their wireless traffic with encryption. This was named Wired Equivalent Privacy (WEP) and it could be activated by the operator of a Wi-Fi network, whether that’s you in your home or the one barista at the coffee shop who knows the admin password. As with any security protocol, WEP spelled out processes for making sure it’s talking to the right device (authentication), verifying that the transmission wasn’t tampered with (data integrity) and protecting the transmission’s privacy (encryption).
After it debuted in the real world, WEP was viciously torn apart by academic researchers and hackers who discovered gap after gap in its security. One of the problems at first was that the size of the cryptographic key used by the underlying 1980s-era RC4 encryption standard was initially a mere 40 bits long; longer keys are stronger, but strong cryptography was still illegal due to its classification as a munition by the U.S. Most of these restrictions were phased out in 2000, allowing longer key sizes, but WEP had other problems. Another flaw included the ability to trick the process which verifies that a chunk of data’s integrity, thus allowing an attacker to modify it en route without any way for you to detect the change. Cryptographic keys were reused, which meant that patterns in how the data was encrypted could be used to break the encryption and read all the data transmitted to and from the device broadcasting a Wi-Fi network’s signal. Eventually, later exploits would allow enough data to be captured in far less time than previous attacks would take, to below 1 minute using a computer of moderate specifications from 2006.
These attacks soon moved from the theoretical realm to the real world. In 2005, hackers broke the WEP encryption used by T.J. Maxx’s cash registers and other store equipment in order to copy employee access data and customer credit card numbers. Some of that captured data then allowed the attackers to connect to T.J. Maxx’s central database, looting 45 million customer records, including usernames and passwords.
WEP was clearly too weak to protect Wi-Fi networks, which were often the primary method people used to access the internet. In 2004, the 802.11 standard was improved to include a new standard which addressed the flaws in WEP, named Wi-Fi Protected Access (WPA). For the most part, WEP was sent to the proverbial glue farm.
A few modifications led to the WPA2-PSK standard (for smaller networks in homes and small businesses) as well as the WPA2-EAP-PSK standard (for larger, managed networks with hundreds or thousands of connected devices). While it hasn’t repeated the same mistakes WEP has, WPA’s underlying architecture and unique characteristics still led to other shenanigans.
Many of the security flaws in Wi-Fi were merely theoretical, laid out in academic papers and not commonly found in the real world. Arguably the most common actual Wi-Fi “attack” does not involve breaking anything. It’s called wardriving, and it’s less like burglary and more like peeking into windows.
If you're old enough to remember analog television, you may have less than fond memories of wiggling an antenna to make the SeaQuest episode you were watching not look like it was actually being transmitted from the bottom of the ocean. The antenna design for many general-purpose wireless access points is usually omnidirectional, means it broadcasts, more or less, in every direction, like a light bulb. Antennas on the receiving end are usually also omnidirectional, but they don’t have to be. Some antenna designs used for beaming Wi-Fi from rooftop to rooftop, for example, are directional, shooting radio waves toward a focused point like a laser beam.
Sometime around 2002, someone figured out you make a directional antenna with a Pringles can.
Now, armed with a handheld antenna and a belly full of chips, one could hop into the passenger seat of a friend’s car and drive around, sniffing out and connecting to open networks. Such cheap, homemade antennas were not as powerful as their expensive commercial equivalents but worked well enough to detect the type of wireless access point, its signal strength, and whether it was an open network or protected with WEP or WPA. Today, using the built-in GPS of your smartphone, you can download an app called Wigle to do the same thing.
Wardriving (or, if one were ambitious enough to acquire more serious vehicles, as technologist Rick Hill did in 2006 and 2008, respectively, “warrocketing” or “warballooning”) was typically practiced by bored tech enthusiasts before they discovered geocaching. Wardriving made it possible to freeload off a neighbor’s internet — you could theoretically do things you’d be afraid to do on your home network, like download porn or pirate movies — but most people simply did it as a hobby, capturing the access point’s location and plotting it on maps online.
Beyond map making, however, wardriving can also work for the reconnaissance of places certain heads of state and their friends hang out in, and find weaknesses that could be exploited to surveil any unencrypted activity. In 2017, Gizmodo’s Surya Mattu warboated Mar-a-Lago, found that it still uses WEP and misconfigured network equipment. While Mattu refrained from actually cracking the WEP key, as it has been possible for over well a decade, it can be assumed that no small number of intelligence agencies would show restraint, and most would have budgets that can cover a boat rental and some Pringles. Beyond surveillance of network traffic, if the aforementioned misconfigured network equipment were misconfigured in just the right way, it would also be possible to try cracking the network’s administrative password to then change the settings of the wireless network, possibly allowing attacks on any poorly-secured network-connected camera systems, adding a visual element to digital eavesdropping.
Although WPA solved WEP’s problems, many public spaces still opt for hosting an open network with no encryption at all. This, along with the still-widespread lack of HTTPS support — a layer of encryption separate from what the network provides between a website and your device’s web browser — created an environment where attackers could do all the fun/terrifying things people theorized about in research papers.
Any hacker who could connect to a Wi-Fi network could potentially cause loads of mischief, ranging from pirating enterprise software from poorly secured shared folders in corporate offices to snooping on an individual’s web browsing. Up until 2010, many websites may have protected their login page with HTTPS but not other important parts of the site which create a “session” — which lets a website remember that you logged in so you don’t have to login every time you click on anything. An attacker could see the pages you visited and possibly read the emails you sent if you were both on the same open Wi-Fi network.
In 2010, software developer Eric Butler released Firesheep, a free downloadable extension for the Firefox browser designed to demonstrate the power of “sidejacking,” which in this case meant capturing the process of creating a website session after logging in and using that to impersonate your account on that site. A Firesheep user could hop onto a network — let’s say, at Starbucks — click “Start Capturing,” and wait as the tool monitored for visits to websites that weren’t fully encrypted with HTTPS — which Facebook was not, at the time Firesheep was released. Once someone else on the network logged in to one of these sites, Firesheep could sidejack their session and use that website as that user without logging in. This was possible because in an open Wi-Fi network with a website that isn’t fully HTTPS encrypted, anyone can record the unencrypted session data relayed through the open air. Facebook, Twitter, Gmail, and Flickr have all enabled HTTPS by default now, which nullifies this attack, and many other sites are headed in that direction. However, there are still some top sites, including Alibaba and IMDB, that do not work over HTTPS, and many more, including the L.A. Times and FiveThirtyEight, that have not moved to HTTPS by default.
Obscure quirks and features of wireless access points lead to other potential problems related to the availability of the network itself. A bit of loosey-gooseyness in the 802.11 specification allowed for the ability to safely disconnect devices from a network, but without strong authentication. Because of this, an attacker only needs to know the hardware identifier known as the “MAC address” of the wireless access point (available to any device connecting to the access point) and the MAC address of the connected device (visible to the attacker in Wi-Fi traffic in an open network). From there, a special command can be sent to the access point to disconnect the device. In other words, an attacker can boot your computer off the network without warning no matter what you’re doing.
As scary as that might sound, the network itself is generally not affected, and most computers and smartphones are set up to reconnect to a network if they’re disconnected. The WPA-EAP protocol designed for larger Wi-Fi networks addresses this, and some networks allow you to prevent this kind of attack, but without the effort put into that configuration, many Wi-Fi networks will remain vulnerable. This kind of attack doesn’t even need the full processing power of a whole computer. You can buy a $30 piece of specialized hardware the size of a AA battery that can automatically launch the attack at the touch of a very tiny button.
WPA2 security was much better than WEP or simply leaving a network unencrypted, and for a while, it seemed to be rock solid. Up until very recently, attacks on WPA2 were variations of “guess the Wi-Fi password” but with fast computers. This is also known as a brute force attack, and can be done by using a program to go through lists of possible passwords or pre-encoded lists of possible passwords known as rainbow tables, which are the same thing but with each possible password pre-encoded the same way the real password is to make it easier for password cracking software to make a comparison. Earlier in October, however, Mathy Vanhoef, an academic researcher at imec-DistriNet Research Group of KU Leuven, discovered a fundamental flaw in WPA2.
The flaw, which was dubbed KRACK, relates to the authentication process for WPA2. Basically, an attacker creates a virtual Wi-Fi access point similar to the one a user is trying to connect to and replays a part of the authentication process to then decrypt a user’s Wi-Fi traffic. The attacker can then start monitoring any unencrypted traffic, more or less as one would in a broken WEP network. Essentially, this drops the security of a Wi-Fi access point back to 2002 levels. If T.J. Maxx is running WPA2, it should probably update the software on its cash registers.
The information security world was initially pretty spooked by the possibility of a protocol-level flaw, but it turned out that the flaw was small enough that device makers could correct for it without the IEEE having to go back to the drawing board and come up with a new Wi-Fi encryption standard. Furthermore, for websites, the attack only works for traffic that passes through the less secure HTTP protocol; data exchanged through sites like The L.A. Times, Idaho State Journal, and 4chan would be accessible, but data passed through Facebook, Amazon, Netflix, Healthcare.gov, or Twitter would not. In this context, KRACK in 2010 would have been spooky, but not so much in 2017.
But while KRACK may not pose much of a problem when it comes to web browsing, the rise of the Internet of Things, which is connecting everything from toasters to cars to the internet, leaves some open questions. IoT device manufacturers tend to be slower to fix vulnerabilities, while the update process on the consumer end is usually more tucked out of sight and easily forgotten about. It’s also difficult to tell whether they separately encrypt their traffic or rely solely on WPA to protect the privacy of the data these devices collect and transmit.
After the KRACK vulnerability was discovered, researchers reflected on the nearly 20 years that Wi-Fi has been around, floating our communications through the air. Even though it’s a common standard that has at times had gaping vulnerabilities, Wi-Fi attacks haven’t been nearly as common or as devastating as other types of hacking.
Ultimately, KRACK, WPA2 brute forcing, and other Wi-Fi attacks all requires an attacker to be in the same physical space as you (or creepily pointing a Pringles can at you from the parking lot outside). Although cracking Wi-Fi security measures can be a stepping stone to accessing other systems, that type of small-batch surveillance is an investment of time and human labor that’s comparatively slow and expensive compared with hacking on the greater internet. For example, something like a botnet — a several-thousand or million-plus-strong collection of hacked computers under the control of a cybercrime cartel — can be rented to execute automated attacks, mine cryptocurrency, or install ransomware on any vulnerable machine on the internet it can find. The internet, being significantly larger than your local Wi-Fi network, holds bigger prizes.
Even targeted surveillance where someone is out to specifically get you reaps bigger gains with a simple “phishing” attack, where someone from anywhere in the world sends a legitimate-looking email linking to a fake login page that captures your username and password — which, if you’re like most people, you also use for everything else you do online. Nation-state actors like the one that got access to John Podesta’s emails use phishing attacks because they’re cheap, scalable, effective, and can have a massive return on investment. Although Wi-Fi attacks aren’t nothing, they’re a tiny campfire next to the flaming dumpster we call internet security.
Correction: Surya Mattu works at Gizmodo, not ProPublica.