Original URL: http://www.theregister.co.uk/2006/09/18/web_vulnerabilties/

Web vulns top security threat index

On the up-and-up

By Robert Lemos

Posted in Security, 18th September 2006 09:02 GMT

Analysis Less rigor in web programming, an increasing variety of software, and restrictions on web security testing have combined to make flaws in web software the most reported security issues this year to date, according to the latest data from the Common Vulnerabilities and Exposures (CVE) project.

A draft report on the latest numbers from the vulnerability database found that 4,375 security issues had so far been cataloged in the first nine months of 2006, just shy of the 4,538 issues documented last year. The data shows that web flaws have continued their meteoric rise since 2005, capturing the top-three spots on the list of most common vulnerabilities. Buffer overflows, a perennial favorite, fell to the No. 4 slot.

"The takeaway is that researchers are paying a lot more attention to web vulnerabilities, and if companies don't want to get caught up in that, then they need to pay attention to those flaws," said Steven Christey, the security researcher that authored the draft report and the CVE Editor for The MITRE Corp., a nonprofit government contractor.

The jump in web-based vulnerabilities is fueled by the simplicity of exploiting many of the most common web vulnerabilities, the enormous number of web applications freely available, and the difficulty in eradicating cross-site scripting flaws. Moreover, while many of the vulnerabilities are easy to test for and find, independent security researchers are less likely to probe another group's website to find the flaws, because doing so violates computer intrusion statutes. The case of Eric McCarty illustrates the danger: The network administrator found a database vulnerability in the online application site for the University of Southern California but was prosecuted for his unauthorized access of the server and last week agreed to plead guilty.

Easy-to-use web programming languages are also to blame, because they attract people who have not programmed before and can be more easily audited for flaws, Christey said.

"The existence of these web-friendly languages, like PHP, lowers the bar for someone to create a useful application but also lowers the bar for someone to find vulnerabilities in that application," he said.

In the CVE Project's latest numbers, flaws that use a technique for injecting code from one website into another, known as cross-site scripting or XSS, accounted for 21.5 per cent of the vulnerabilities reported so far in 2006.

Cross-site scripting is considered by many security researchers to be a less-than-hackerly technique used by script kiddies, phishers and spammers to fool trusting users. The technique is a key method for injecting malicious code into a victim's web session. Cross-site scripting allows a malicious website to inject code into the context of another website; a user that believes they are interacting with a popular social networking site, for example, might instead be loading a script in from some other malicious site.

In separate presentations at the Black Hat Security Briefings in Las Vegas in August, two researchers warned that such flaws, paired with high-level scripting languages such as Javascript and AJAX, could be used by data thieves and online attacker to compromise systems using a web worm. Last year, the Samy worm used just such techniques to propagate among MySpace users, using Javascript and AJAX code to add a MySpace.com user "Samy" to the victim's list of friends.

Data from the CVE Project also highlighted two other types of common web vulnerabilities. Database injection vulnerabilities, which accounted for 14 per cent of the flaw reported to date in 2006, are security holes that allow an outside user to inject commands into the database powering a given site. PHP remote file inclusion vulnerabilities, which accounted for 9.5 per cent of flaws so far this year, occur in applications created with the popular PHP web programming language and which allow a malicious user to load their own PHP program to the remote website. The two classes of flaws ranked second and third, respectively, on the CVE Project's list of common flaws.

The difficulty in eradicating cross-site scripting flaws will likely mean that online attackers will continue to focus on the vulnerabilities and find better ways of exploiting them, said Brian Chess, chief scientist with code-checking tool maker Fortify Software.

"With cross-site scripting, it's hard to write good code to exploit the flaws - I think that will change," Chess said. "But doesn't this all sound familiar? It sounds like the same thing as what we were seeing with buffer overflows."

Buffer overflows have been a thorn in the side of security-conscious programmers for more than three decades. While cross-site scripting flaws might not last that long, they have all the makings of a bad headache for security professionals, MITRE's Christey said.

"Their potential for high impact security incidents is not yet fully explored," he said. "Look to the MySpace worm and think about the sort of damage that could have been done in the hands of a malicious individual."

With greater interconnectivity promised by such technologies as Web 2.0, network administrators may have to look forward to just such threats in the future.

This article originally appeared in Security Focus.

Copyright © 2006, SecurityFocus