Outsourced software is a hotspot for the sort of hidden security problems that leave applications vulnerable, an analysis by code testing outfit Veracode has warned.
The company's latest 'State of Software' analysis looked at nearly 3,000 applications submitted by third-parties for security code review, finding that 57 percent failed to meet acceptable levels for security-worthiness on first pass.
In terms of exposure to the OWASP (Open Web Application Security Project) Top 10 most critical web application flaws, only 40 per cent made the grade.
Drilling down by development sector, Veracode found that outsourced software code posed the biggest security risk, with a staggering 93 per cent of apps looked at failing to reach an acceptable standard on first assessment.
Although the numbers involved are small - the company only assessed around 30 outsourced apps out of a total of 2,922 - the issue of outsourced code vulnerability is not a new worry. In 2008, recently acquired rival Fortify Software found that outsourced code was a major area of unknown coding behaviour which could lead to problems.
Up to 76 per cent of internal code had elements of outsourced of third-party code within it, so some of the problem could be hidden within this sector.
"It indicates the amount of security testing and education that has gone on," said Veracode CTO, Chris Wysopal of outsourced app houses.
Open source meanwhile scored above other development sectors in terms of the OWASP Top 10 most critical web application flaws, with 49 per cent of open source apps passing this test on pass one, against 12 per cent of internally-developed apps, and only 7 per cent of commercially-developed apps.
For general security flaws, 58 per cent of open source apps failed on first inspection, roughly on par with the 54 per cent figure for internal apps, but better than the 66 per cent of commercial apps that failed at this stage.
In Wysopal's view, some of this differences between sectors might be explainable by the fact that the commercial apps submitted were more likely to be interactive apps inherently more prone to security problems.
Internal apps, by contrast, tended to be higher quality for a simple reason. "They [the development team] are going to run the code they write. They have a vested interest in security." Open source also did respectably because of the process of open code review that was inherent to its development model.
Once security issues had been identified, open source remediated the problems in an average of 12 days, internal software in 15 days, and commercial software in 19 days.
Fifty-two percent of the vulnerabilities that needed remediation in static analysis were found to be cross-site scripting related, with .NET showing an unusually high frequency of this issue. Java was another soft spot.