Hackers are actively using exploit code to target a flaw in Microsoft software that generated a special warningfrom the US government last week.

The problem concerns a networking function called Windows Server services within the Windows OS, used for file sharing and printing.

Microsoft issued a patch, MS06-040, MS06-040, for the problem on 8 August, which affected several Microsoft OSes. Security experts warned then that exploit code had been detected and could be used more widely.

However, the latest exploit code only affects users running Windows 2000 who have not applied the patch, Microsoft said. The impact so far from the malware, which the company calls "Win32/Graweg," has been low, the company said.

"We are not currently aware of widespread customer impact," said Microsoft.

The SANS Institute reported on Sunday other names given to the malware by security vendors. Symantec calls it "W32.Wargbot" while TrendMicro has named it "Worm.IRCBOT.JK and JL." McAfee goes by "IRC.Mocbot" and F-Secure refers to the malware as "IRCBOT-ST."

The malware is a "bot," a class of malicious code that allows a hacker to take remote control over a computer. The malware appears to be a version of one called "Mocbot," which first appeared in late 2005, according to Lurhq, a security company. Both SANS and Lurhq said two similar versions of the bot are circulating.

Once on an infected machine, the bot contacts remote servers in China over IRC, Lurhq said.

"Historically, Chinese ISPs and government entities have been less-than-cooperative in taking action against malware hosted and controlled from within their networks," said Lurhq.

The bot is capable of several malicious functions, Lurhq said. It can send messages through a user's AOL IM account, an activity that could be used to trick other users into downloading the bot. It's also possible to use the bot to launch a DDoS attack, Lurhq said.

The bot could also spread itself to other computers on a network, giving it worm-like characteristics. Microsoft, however, said the exploit code does not appear to be self-replicating at this point.