A U.K. analytics firm that warned earlier this week of an information leak in Internet Explorer (IE) today rebuked Microsoft for downplaying the bug.
Microsoft, however, has announced it is working on a fix, although the nature of the anticipated patch was unclear.
Criminals could use the technique, Spider.io alleged, to monitor mouse movements used to log into sensitive websites with "virtual keyboards," on-screen keyboards similar to those on smartphones. Some websites, notably a few banking sites, rely on virtual keyboards as a way to stymie the far-more-common malware that captures keystrokes from a physical keyboard.
Yesterday, Microsoft downplayed the threat, noting -- as Spider.io had also charged -- that only a pair of advertising analytics companies have taken advantage of the bug. Those firms, said Spider.io, monitor cursor movement to track whether an ad is visible to the user, or whether it is hidden because the web page is larger than the viewing area of the browser.
Spider.io relies on its own technology to determine what proportion of the ad is visible. The technique, which Spider.io CEO Douglas de Jager labeled "browser optimization" in an October interview, watches how a browser allocates resources to render an ad.
The U.K. firm posted a message on Bugtraq, one of the most popular security mailing lists, on Tuesday, then followed that with a blog post Wednesday, recounting how it reported the bug to Microsoft on Oct. 1, but was later told by Microsoft that the Redmond, Wash. developer had no plans to fix the flaw.
Microsoft's top executive for IE, Dean Hachamovitch, took to a company blog Thursday to counter Spider.io's claims.
Hachamovitch downplayed the threat to Windows users, and said that Spider.io's report was motivated more by spite than by concerns over security. "From what we know now, the underlying issue has more to do with competition between analytics companies than consumer safety or privacy," said Hachamovitch. "The only reported active use of this behavior involves competitors to Spider.io providing analytics."
However, Hachamovitch did promise that Microsoft was on the case. "We are actively working to adjust this behavior in IE," he said.
The "adjustment," he said, would bring IE into line with other browsers' behavior. "Analytics firms can expect to do viewpoint detection in IE similarly to how they do this in other browsers."
Ad visibility can be monitored in other browsers, such as Chrome and Safari, with different techniques, but those work only when the tracking code has been placed on the same page as the ad, and when the ad originates from the same domain as the Web page's content.
In other words, ad analytics companies would not be able to "game" IE to get more information than they would get from a competing browser.
In a Friday rebuttal of Hachamovitch, de Jager added more to the debate. He did not address Hachamovitch's allegation that competition drove the bug's disclosure, however, or respond to Computerworld's questions on the topic.
Instead, de Jager took Microsoft to task for refusing to address the problem, even though that had been made moot by Hachamovitch's statement that Microsoft is working on a fix of some kind.
"It isn't for Microsoft and the various companies currently exploiting the vulnerability to decree unilaterally that this vulnerability is not important enough to fix," said de Jager. "According to existing privacy standards, it is not OK for a browser to leak your mouse coordinates outside of the particular browser window." De Jager suggested that the decision should be put in the hands of privacy experts.
He also blasted Hachamovitch for minimizing the threat with language such as "theoretical" and "very little risk to consumers."
"Ads do not need to be served to sites requiring login details," said de Jager. "Ads need only to be served to some page which is open in Internet Explorer."
Hachamovitch had pointed out that there was no reliable way for an attacker to know what was beneath the mouse cursor at any given point, and so criminals would be unlikely to figure out which mouse movements were, for instance, related to a virtual keyboard. At one point, Hachamovitch said it was "hard to imagine" how attackers could put all the necessary pieces together to pull off the theft of virtual keyboard entries.
De Jager did not address that aspect of Hachamovitch's rebuttal, or reply to questions about how the IE bug might be exploited in real-world attacks.
Gregg Keizer covers Microsoft, security issues, Apple, Web browsers and general technology breaking news for Computerworld. Follow Gregg on Twitter at @gkeizer, on Google+ or subscribe to Gregg's RSS feed . His email address is firstname.lastname@example.org.