WCry is so mean Microsoft issues patch for 3 unsupported Windows versions

A day after a ransomware worm infected 75,000 machines in 100 countries, Microsoft is taking the highly unusual step of issuing patches that immunize Windows XP, 8, and Server 2003, operating systems the company stopped supporting as many as three years ago.

The chaos surprised many security watchers because Microsoft issued an update in March that patched the underlying vulnerability in Windows 7 and most other supported versions of Windows. (Windows 10 was never vulnerable.) Friday's events made it clear that enough unpatched systems exist to cause significant outbreaks that could happen again in the coming days or months. In a blog post published late Friday night, Microsoft officials wrote:

We also know that some of our customers are running versions of Windows that no longer receive mainstream support. That means those customers will not have received the above mentioned Security Update released in March. Given the potential impact to customers and their businesses, we made the decision to make the Security Update for platforms in custom support only, Windows XP, Windows 8, and Windows Server 2003, broadly available for download here.

This decision was made based on an assessment of this situation, with the principle of protecting our customer ecosystem overall, firmly in mind.

Crucial entry point still missing

Microsoft announced the patches around the same time it said it still doesn't know what the precise starting point was for Friday's WCry outbreak. One of the key questions circulating once Friday's viral outbreak appeared to be contained was how did the self-replicating worm first gain entry so it could go on to spread from vulnerable machine to vulnerable machine.

At least two security firms—FOX-IT here and CrowdStrike here—said spam that sent fake invoices to end users provided the crucial initial vector to seed the self-replicating attack, but none of the three companies have produced copies. Some researchers doubted a generic e-mail campaign could have been the sole initial vector without leaving a mountain of evidence that would have surfaced by now. In a blog post published Friday night, Microsoft officials wrote:

We haven't found evidence of the exact initial entry vector used by this threat, but there are two scenarios we believe are highly possible for this ransomware family:

• Arrival through social engineering emails designed to trick users to run the malware and activate the worm-spreading functionality with the SMB exploit
• Infection through SMB exploit when an unpatched computer can be addressed in other infected machines

The blog post went on to say that the worm "executes massive scanning on Internet IP addresses to find and infect other vulnerable computers."

FOX-IT also said in its blog post that "there appear to be multiple infection vectors," but the post didn't elaborate. Maarten van Dantzig, a researcher with FOX-IT, said on Twitter here and here that he suspects e-mail was the initial vector for some, but not all, of the outbreaks. Researchers from Cisco Systems Talos group went even further, writing: "Our research does not yet support that e-mail was the initial infection vector. Analysis is ongoing."

The possibility that ransomware can spread virally across the Internet without any form of end-user interaction is a chilling prospect. Internet-wide scans performed in recent weeks show that as many as 2.3 million computers have the necessary port 445 exposed to the Internet. Those scans also reveal that 1.3 million Windows machines haven't been patched.

People who are running unpatched machines should take action immediately. The best measure is to patch the vulnerability using this link for supported versions or this one for XP, 8, and Server 2003. Those who can't patch should ensure their computers are locked down by, among other things, blocking outside access to ports 138, 139, and 445. They should also disable version 1 of the Server Message Block protocol.

Friday's attack could have been much worse, had the perpetrators not slipped up by failing to register an Internet domain that was hardcoded into their exploit as a sort of "kill switch" they could activate if they wanted to shut down the worm. That made it possible for a quick-acting researcher to register the domain and stop much of the attack just as it was gaining momentum.

A new attack could come at any time. Next time, defenders may not be so lucky. As Microsoft's blog posts makes clear, vulnerable machines aren't only a danger to themselves, but to the entire world at large.

Listing image by John Lodder