This feels like it could be an attack vector. Gather intel on what the user agent is, nmap the IP, possibly find a vulnerability in the parser or the server.
You'd think... but after that story about Microsoft just executing random threatening code it found on someone's computer and allowing it access to the internet, I have to question some of the wisdom these big companies show.
That's assuming Microsoft didn't do things properly. Who's to say the amount of connections that could be opened, the bandwidth, or the max traffic that could be recv/sent wasn't limited?
Thinking that you can make an HTTP request using this method and that that means you can unleash a DoS is... worth a try, but not something you can take for granted.
