Risk
9/30/2010
07:20 PM
George V. Hulme
George V. Hulme
Commentary
50%
50%

In Software We (Can't) Trust

I can't think of more than a few attacks in the past decade that involved stolen certificates as part of the malware or exploit code. However, recent attacks, and new research highlights the increasing danger of trusting signed digital certificates.

I can't think of more than a few attacks in the past decade that involved stolen certificates as part of the malware or exploit code. However, recent attacks, and new research highlights the increasing danger of trusting signed digital certificates.While the danger of spoofed and forged certificates had always been a theoretical concern, it was Microsoft's spoofed certificates in 2001 when I first became aware of real-world attacks like this.

Today, they are growing commonplace.

Consider the Adobe attacks against Reader and Acrobat that exploit flaws as well as stolen certificates. from DarkReading earlier this month:

Meanwhile, Roel Schouwenberg, senior antivirus researcher for Kaspersky Lab, studied an attack exploiting the flaw that uses a stolen digital certificate from a credit union to sign the infected PDF file -- akin to what the Stuxnet attacks did. Schouwenberg says as this technique takes off, it will result in more missed attacks as well as more false positives from security software. "I predict that the security industry will have more misses of these files that come with stolen signatures, or [have] more false positives. We could well be in this high false positives [trend] next year, which we haven't seen in a while," he says.

As Schouwenberg points out, Stuxnet also used stolen digital certificates to do its thing. The worm that allegedly targets Iran's nuclear program used two stolen digital certificates from two separate Taiwanese technology firms.

Mike Wood, researcher at anti-virus firm Sophos, gave a talk today at this years' Virus Bulletin conference (wish I was there) about the abuse of digital signatures to increase the reputation of fraudulent software, or as part of how malware protects itself.

Wood's paper, which won't be widely available until after the conference, 'Want My Autograph?': The Use and Abuse of Digital Signatures By Malware, is an interesting read on Microsoft's Authenticode Program, which requires Windows kernel-mode software 'drivers' to be digitally signed - and how they can be abused. The paper also details Web-based PKI abuse, phishing attacks with certificates as bait, and many other challenges associated with digital certificates.

Wood concludes (unsurprisingly) that anti-virus software is well positioned to fight malicious or stolen certificates. I'm sure anti-virus has its place, but much more can be done by the software companies and the Certificate Authorities that sell these certificates to ensure those in use are legitimate.

In that story (also linked above) I wrote in 2001 about the spoofed Microsoft certificates, the analyst I interviewed came to a conclusion that is just as true today as it was nearly a decade ago:

Analysts say this incident shouldn't take away from the strengths of digital certificates as a security tool, but it does point to the weakness of the digital-certificate-assignment process. Says Hurwitz Group analyst Pete Lindstrom, the initial authentication process is "the Achilles heel" of public key infrastructure.

Woods' paper, unfortunately, isn't available until after the Virus Bulletin conference is over, so I'll update this post with a link as soon as it becomes available.

For my security and technology observations throughout the day, find me on Twitter.

Comment  | 
Print  | 
More Insights
Register for Dark Reading Newsletters
White Papers
Video
Cartoon
Current Issue
Flash Poll
Slideshows
Twitter Feed
Dark Reading - Bug Report
Bug Report
Enterprise Vulnerabilities
From DHS/US-CERT's National Vulnerability Database
CVE-2014-2329
Published: 2015-08-31
Multiple cross-site scripting (XSS) vulnerabilities in Check_MK before 1.2.2p3 and 1.2.3x before 1.2.3i5 allow remote authenticated users to inject arbitrary web script or HTML via the (1) agent string for a check_mk agent, a (2) crafted request to a monitored host, which is not properly handled by ...

CVE-2014-2330
Published: 2015-08-31
Multiple cross-site request forgery (CSRF) vulnerabilities in the Multisite GUI in Check_MK before 1.2.5i2 allow remote attackers to hijack the authentication of users for requests that (1) upload arbitrary snapshots, (2) delete arbitrary files, or possibly have other unspecified impact via unknown ...

CVE-2014-2331
Published: 2015-08-31
Check_MK 1.2.2p2, 1.2.2p3, and 1.2.3i5 allows remote authenticated users to execute arbitrary Python code via a crafted rules.mk file in a snapshot. NOTE: this can be exploited by remote attackers by leveraging CVE-2014-2330.

CVE-2014-2332
Published: 2015-08-31
Check_MK before 1.2.2p3 and 1.2.3x before 1.2.3i5 allows remote authenticated users to delete arbitrary files via a request to an unspecified link, related to "Insecure Direct Object References." NOTE: this can be exploited by remote attackers by leveraging CVE-2014-2330.

CVE-2014-2570
Published: 2015-08-31
Cross-site scripting (XSS) vulnerability in www/make_subset.php in PHP Font Lib before 0.3.1 allows remote attackers to inject arbitrary web script or HTML via the name parameter.

Dark Reading Radio
Archived Dark Reading Radio
Another Black Hat is in the books and Dark Reading was there. Join the editors as they share their top stories, biggest lessons, and best conversations from the premier security conference.