Google released a report on ClickBot.A, a click fraud bot net.

Here's the page

http://adwords.blogspot.com/2007/04/new-case-study-on-botnet-based-click.html Google announcement

Here's the pdf

http://www.usenix.org/events/hotbots07/tech/full_papers/daswani/daswani.pdf [PDF]

Google notes all the clicks from the 'bots – $50k worth – were invalidated and thus advertisers weren't charged. All well and good.

The botnet was detected by Panda Software, the virus folks.

What isn't clear is when Google noticed the bad clicks – was it before the network tripped up filters at Panda? If so, Google found these criminals just via click signatures. Or was it after Panda detected the Trojan and the bot-master site? If the later, Google's click analytics hadn't found the crime, due to its low frequency (max 20 clicks per bot, 15 mins apart, but spread across 100k bots)

More musings on the advertising implications of this here:

http://www.rimmkaufman.com/rkgblog/2007/04/11/follow-the-clickbota-money-trail/