Bug hunters, software firms in uneasy alliance

Tom Ferris is walking a fine line. He could be Microsoft's friend or foe.

Ferris, an independent security researcher in Mission Viejo, Calif., found what he calls a serious vulnerability in Microsoft's Internet Explorer Web browser. He reported it to the software giant on Aug. 14 via the "secure@microsoft.com" e-mail address and has since exchanged several e-mail messages with a Microsoft researcher.

Up to that point, Ferris did everything according to Microsoft's "responsible disclosure" guidelines, which call for bug hunters to delay the announcement of security holes until some time after the company has provided a fix. That way, people who use flawed products are protected from attack, the argument goes.

Last weekend, however, Ferris came close to running afoul of those guidelines by posting a brief description of the bug on his Security Protocols Web site and talking to the media about the flaw. So far, the move has done little more than raise some eyebrows at Microsoft.

"I am walking a fine line, but I am doing it very carefully because I am not disclosing actual vulnerability details," Ferris said. "I do this to inform users that flaws still do exist in IE...I don't like it that Microsoft tries to give users a nice warm feeling that they are disclosing everything researchers report to them."

At issue is the push for "responsible disclosure" of software flaws by many industry players, including titans such as Microsoft, Oracle and Cisco Systems.

Microsoft publicly chastises security researchers who don't follow its rules. Also, those researchers won't get credit for their flaw discovery in Microsoft's security bulletin, which is published when the company releases a patch. Because Ferris did not disclose any actual vulnerability details, he's still on Microsoft's good side, a company representative said.

While many software makers promote responsible disclosure, it isn't universally backed by the security community. Critics say it could make security companies lazy in patching. Full disclosure of flaws is better, they say, and turns up the heat on software makers to protect their customers as soon as possible.

How long is too long?
"Microsoft obviously takes way too long to fix flaws," Ferris said. "All researchers should follow responsible disclosure guidelines, but if a vendor like Microsoft takes six months to a year to fix a flaw, a researcher has every right to release the details."

By that time someone else, perhaps a malicious person, may also have found the same flaw and might be using it to attack users, Ferris said.

Often lambasted for bugs in its products, Microsoft is doing its best to win the respect of the security community. The company has "community outreach experts" who travel the world to meet with security researchers, hosts parties at security events and plans to host twice-annual "Blue Hat" events with hackers on it its Redmond, Wash., campus. At Blue Hat, hackers are invited to Microsoft's headquarters to demonstrate flaws in Microsoft's product security.

"Security researchers provide a valuable service to our customers in helping us to secure our products," said Stephen Toulouse, a program manager in Microsoft's security group. "We want to get face to face with them to talk about their views on security, our views on security, and see how best we can meet to protect customers."

Many companies are getting better at dealing with security researchers, said Michael Sutton, director of iDefense Labs, which deals with researchers and software makers. "The environment has definitely changed from two or three years ago, though there are vendors who are going in the opposite direction," he said.

While Microsoft sometimes is still referred to as the "evil empire," it appears to be successfully wooing security researchers.

"We are at the point where all the obvious things we tell Microsoft to do, they already do it," Dan Kaminsky, a security researcher who participated in Microsoft's first Blue Hat event last March, has said.

Balancing act
Other technology companies still struggle with hacker community relations. Cisco especially has managed to alienate itself from the hacker community to the extent that T-shirts with anti-Cisco slogans were selling well at this year's Defcon event. Oracle also isn't a favorite, researchers said.

Cisco, along with Internet Security Systems, last month sued security researcher Michael Lynn after he gave a presentation on hacking router software at the Black Hat security conference. The company had previously tried to stop Lynn from giving his talk in the first place.

"It was definitely a surprise to see Cisco's reaction," iDefense's Sutton said. "I don't think

that's the best approach. I do feel that it is happening less and that vendors are realizing that we don't want to work against them, but with them."

Cisco contends it doesn't have any beef with Lynn's discoveries, but instead the company is unhappy about the way he went about distributing the information to the public.

"This incident violated aspects of normal protocol for dealing with security flaws," said Bob Gleichauf, CTO for Cisco's Security Technology Group. "And we are real sticklers for protocol."

But it seems that there have been several instances where Cisco has had similar problems in its dealings with researchers.

Early in 2004, Paul Watson discovered a flaw in the TCP/IP protocol that could be exploited on a number of networking products, including Cisco's routers. Watson said he initially e-mailed two of Cisco's engineers, who responded promptly. They were helpful and even contributed some thoughts and ideas to his research, he said.

But once the issue was identified as a serious security risk by the legal team at Cisco, the tone of the communication changed, Watson said. Cisco still wanted information from Watson, but no longer responded to his queries. Watson provided Cisco with several possible methods to correct the problem.

Frustrated by the lack of communication with Cisco, Watson decided to present his research at the CanSecWest Security Conference in April 2004. In a scenario similar to that at Black Hat, Cisco and the U.S. Department of Homeland Security asked the conference organizer to pull the talk. The request was denied.

The impending talk spurred the company into action. Fixes were released a few days before the conference. However, Cisco not only provided patches, it also patented a fix for the flaw. This raised fears that Cisco might charge for the fix, which also affected other vendors, although Cisco did not.

"I was shocked," Watson said in an e-mail. "It really broke my trust in them." Cisco, like other software makers, wants security researchers to report flaws privately and have time to patch before disclosure, but Cisco took advantage of this period to apply for a patent, he said.

Playing it smart
A similar situation played out about a year later. Cisco tried to patent a fix to a flaw in the ICMP protocol that was discovered by Fernando Gont. The researcher outsmarted Cisco by documenting his discovery and the fix, and also by sharing the information privately with the open-source community and the Internet Engineering Task Force, a standards organization.

Mary Ann Davidson, chief security officer at Oracle, sees security researchers who threaten vendors with disclosure of bugs as a problem, she wrote in a recent perspective piece on News.com. "The reality is that most vendors are trying to do better in vulnerability handling. Most don't need threats to do so," Davidson said.

Alexander Kornbrust specializes in security of Oracle products. He went public with details on six security vulnerabilities in Oracle software in July, about two years after he reported the bugs to the software maker and fixes still had not been provided.

Oracle chided Kornbrust as irresponsible for disclosing the data.

Although not entirely happy about his dealings with Oracle, Kornbrust said it is not an adversarial relationship. "Hostile is not the right expression. I did get feedback from Oracle," Kornbrust said. But that was only immediately after he reported the bugs. Oracle did not give Kornbrust updates on how it was addressing the problems afterwards.

"Oracle supports guidelines for responsible disclosure. One of those guidelines is that the company should send out updates to the researcher. They don't," said Kornbrust, who runs Germany's Red Database Security.

In the past, many hackers and security researchers outed glitches without giving much thought to the impact the disclosures would have on Internet users. Software makers have been working to provide a channel for disclosure. Several have also established patching schedules. Microsoft releases patches every second Tuesday of the month, and Oracle has a quarterly schedule.

Still, the debate on responsible disclosure rages. Recently the French Security Incident Response Team, or FrSIRT, was the subject of discussion on a popular security mailing list. FrSIRT, formerly known as K-Otic, releases details on vulnerabilities and also publishes exploit code that could help attackers. Sometimes the holes aren't yet patched. Other than FrSIRT selling its service, what good can such publishing do? critics have asked.

"With our dependency on IT systems, responsible disclosure is of paramount importance," said Howard Schmidt, an independent security consultant who has served as cybersecurity adviser to the White House and security executive at Microsoft and eBay.

Technology companies that are not responsive to security researchers do pose a problem, Schmidt said. He suggests that the government, specifically the US Computer Emergency Readiness Team (the Department of Homeland Security's Internet security agency), could act as an intermediary. "And then perhaps the government could put some pressure on (technology companies)," he said.