Slashdot Mirror
Wi-Fi WPA2 Vulnerability Found
BobB-nw sends along news based on yet another press release in advance of the Black Hat conference: a claimed vulnerability in WPA2 Enterprise that leaves traffic open to a malicious insider. "...wireless security researchers say they have uncovered a vulnerability in the WPA2 security protocol, which is the strongest form of Wi-Fi encryption and authentication currently standardized and available. Malicious insiders can exploit the vulnerability, named 'Hole 196' by the researcher who discovered it at wireless security company AirTight Networks. The moniker refers to the page of the IEEE 802.11 Standard (Revision, 2007) on which the vulnerability is buried. Hole 196 lends itself to man-in-the-middle-style exploits, whereby an internal, authorized Wi-Fi user can decrypt, over the air, the private data of others, inject malicious traffic into the network, and compromise other authorized devices using open source software, according to AirTight. 'There's nothing in the standard to upgrade to in order to patch or fix the hole,' says Kaustubh Phanse, AirTight's wireless architect who describes Hole 196 as a 'zero-day vulnerability that creates a window of opportunity' for exploitation." Wi-Fi Net News has some more detail and speculation.
so rather than a hole, its more a forced proxy? a user who knows your password, is decrypting your traffic, and re-broadcasting it with different content... if this user has your password, you need to have a think about who you give your password to
portfolio
You have an awfully low UID for such a huge troll!
This vulnerability is only useful if the attacker knows your WPA key. In other related news, it has been discovered that those who know your root password can delete all your files.
In other news, people on your wired ethernet segment can also see your "private" traffic. If you care so much, use SSL. Next scaremongering non-story in 3, 2, 1.
"I'm starting with the man in the middle
I'm asking him to change his ways
Every packet is encrypted just a little
If you wanna make your network a safer place
Find the man in the middle and punch his face."
There is an out-of-band key exchange. It is called a trusted certificate. You know, just like how HTTPS works. This is for WPA2 Enterprise, of which there are many different EAP methods possible, but for which most do include an out of band key exchange (i.e., certificates, or EAP-FAST PAK). In any case, there's also the old DH key exchange, which worked fine for IPsec for years.
nah, things went downhill about the 50k mark... ;)
I've been telling people to use VPN over WiFi connections forever. Even better, put your wireless devices on the outside of the firewall, so they have no choice but to VPN in. This also makes giving a random guest access to your wireless no big deal. Any one who thinks wireless networking will ever be safer than an old-fashioned hub is deluding themselves.
I'd say more around the 5170-mark, myself.
Doubting the existence of evolution is like doubting the existence of China: It just shows that you're uninformed.
nah, things went downhill about the 50k mark... ;)
Not really. Things went downhill much sooner than that. I'd have a much lower UID than I have if I had seen the need for it, but the 'first poster' morons, etc., weren't much yet around, and there wasn't much value to HAVING a Slashdot account until some time after the account system was first implemented.
The whole point of a switch is that it sends data only to the host that it is for. So you don't get my data out your switch port. If you clone a MAC, that doesn't do the trick as it just confuses the switch and some data goes to one computer, some to the other, and the connection works poorly. Back in the day you could overload the switches in various ways and make them act like hubs, but that is also noticeable, and it doesn't work on new high quality switches.
Wired networks are actually pretty secure from snooping over all. It's not impossible, but it is damn hard.
Statements like, "I could break any WiFi in about two hours," are red flags that you should higher a different security researcher...
The terms "any", "ever" or "all" are not in most security researcher's vocabularies when talking about unknowns or speculative situations.
We prefer to use terms that imply some degree of uncertainty such as "mostly", "almost never", and "nearly all" since the one thing we know
as security researchers is "trust no one", followed closely by "there is almost always an exception to the rule".
I'm certain that there is at least one "WiFi" your researcher could not break in approximately two hours, thus voiding the "any" term they used.
When in doubt just say, "Prove It."
...I'm using WEP, so I am perfectly safe!
My first program:
Hell Segmentation fault
Because in practice, making sure that there is absolutely no hint of a secure piece of information is incredibly tricky. Most programmers traditionally have little concept of actual *secure* programming. Most implementations of perfectly secure algorithms are subject to flaws because people didn't treat side-cases, or properly analyse how the traffic use would affect the algorithm, etc. e.g. not renegotiating keys often enough, so that people can see enough traffic to decrypt a key in a relatively short space of time.
Additionally, this isn't an attack on the crypto. The crypto secures the conversation, it does not necessarily prove identity and if it does prove identity most places don't care about the identity (how many company distinguish individual users/computers over the wireless network by anything other than MAC/IP/username given? AES is still 100% perfectly intact. If you'd been using, say, OpenVPN or OpenSSH with the same algorithm over an unsecured wireless network, the internal encrypted conversation would still be virtually as secure today as it was when AES was invented. The problem is that the *implementation* of AES wasn't designed to cover the usage scenario here, and probably never could be because of the way the access to this particular tiny piece of this part of the broadcast specification is granted. Basically, the flaw has always been sitting there in WPA, not in AES which is still chugging along nicely doing its job. Shocking that a wireless "encryption" fails to properly implement a security scheme because of a bad implementation that side-steps the actual encryption itself... that's never ever happened before ever anywhere :-P
Moral of the story: only trust crypto from those well-established in the crypto-field that's been attacked and attacked and still is approved for government/military use in lots of sensible countries. And then make sure you have a damn good implementation that's not overly complex, or cast in stone, such that most people can't examine it / play with it / fix it.
If you'd been running OpenVPN over the same wireless network, but using OpenVPN's key infrastructure and encryption instead of WPA or WEP or anything at all (i.e. completely "open" wireless) you would still be secure. A bad implementation of a particular encryption in WPA allows people to bypass steps of the actual encryption process that were never designed to be bypassed. It's almost an "out of band" security vulnerability - i.e. nothing to do with whether you use AES or Blowfish or 3DES or whatever you choose... they basically find a way around the (still theoretically secure) encryption that has no effect on the efficacy of the encryption itself.
Basic rule: Just because your "Ethernet-over-the-mains" devices says it uses AES, don't think that means it's "secure". Chances are that it's not.