If there's malware on the machine, and there is a connected USB token, then authentication is only as good as the password - malware can probe the connected token as often as desired. And this data stream to the authentication host is still subject to a variety of MITM attacks. In the event of an unconnected OTP token, a variety of MITM attacks still applies to OTP tokens - in the SecurID-style form factor, printed lists or anything similar. In theory, with trusted data paths everywhere (internal to worksation as well as he network) OTP is better than passwords alone. But since this data patch assumption is rarely 100% valid, OTP is as good as a password alone. In the situation where data paths are trust-able, OTP is a somewhat better than passwords alone. Does the risk justify the costs involved (tokens, token management, authentication host, and trusted data paths)? Lyal -----Original Message----- From: full-disclosure-bounces@xxxxxxxxxxxxxxxxx [mailto:full-disclosure-bounces@xxxxxxxxxxxxxxxxx] On Behalf Of Bojan Zdrnja Sent: Sunday, 10 September 2006 8:51 AM To: 3APA3A Cc: full-disclosure@xxxxxxxxxxxxxxxxx; bugtraq@xxxxxxxxxxxxxxxxx Subject: [Full-disclosure] Re: RSA SecurID SID800 Token vulnerable by design On 9/9/06, 3APA3A <3APA3A@xxxxxxxxxxxxxxxx> wrote: > Dear Hadmut Danisch, > > 2-factor authentication is not a way to protect against malware. Well, it protects - the authentication process. > SecurID authentication supports single sign-on technology. As a > weak side of this technology, it means, if single account on any > network host is compromised, this account is compromised in > whole network, because any resource can be accessed from compromised > host. An ability to read current key from device is required to > support single sign-on. It depends on the underlying SSO technology. In most cases today you have web based SSO deployments which rely on a cookie. In this case, you don't need to connect the token at all - all you have to do is login once and the browser will take care of rest. As Brian noted in the following e-mail, if an attacker can put a keylogger on your machine, he can certainly get the cookie as well and use it. > The only additional attack factor this issue creates is attacker > can get _physical_ access to console with user's credentials _any > time_ while user is logged in, while in case token can not be red > (e.g. it's not plugged to USB) he can only access console short after > user logs in to compromised host (while token is not changed). No - the OTP can be used only once, so even if you manage to get both the PIN/password and the OTP (remember, you need both to login) you can't use that because the RSA authentication manager (the server side of the whole process) marked that OTP as used. In this case an attacker can only try to brute force the OTP (after all, it's only 6 digits), but RSA has excellent measures against brute force attacks (basically, after a certain, configurable, number of unsuccessful logins the token is disabled; what's even better is that it tracks number of incorrect OTPs with correct PINs - if that is higher than a certain number, it puts the token into "2nd OTP mode" which means you have to guess 2 OTPs in a row). I think these tokens offer excellent means for authentication. Sure, they are not a silver bullet and don't solve all your security problems (nothing does), but if you have users who have to login from a lot of insecure places (airport lounges, cyber caffes) and are afraid of keyloggers stealing passwords, two factor authentication really helps. Cheers, Bojan _______________________________________________ Full-Disclosure - We believe in it. Charter: http://lists.grok.org.uk/full-disclosure-charter.html Hosted and sponsored by Secunia - http://secunia.com/
