AdWords now has fraud analysis accessible to advertisers
Here's an interesting article from the Google AdWords team that touches lightly on the serious issue of clickfraud:
Estimating Invalid Clicks in AdWords
Clickfraud, in case you haven't bumped into the term before, is the bane of online advertising when you're paying for clicks that aren't from potential customers but rather people who are trying to inflate their own click thru rates to increase their AdSense payout. It's easy to understand why publishers might have the urge to tell their friends "click on my links, I'll make more money", but from an advertisers perspective, it's terrible because it's ripping you off for your hard-earned marketing budget.
Search Engine News
I believe Google is misleading advertisers into thinking that it has
click fraud under control. What they call invalid clicks are
situations that can occur under normal circumstances (but also under
fraudulent ones) that they don't charge for. However, they can't
determine whether or not someone is clicking with the intent of
committing fraud. (No one can). It's easy to pick out repeated
visits from the same IP address that run up an advertiser's budget,
but virtually impossible to distinguish between interested parties
clicking on ads and a set of compromised computers or a (human) click
ring doing the same.
There's a new kind of Adwords fraud: account fraud. My account was hacked and a phony campaign was created running up $20,000 in charges in one day! They have no AI in place to detect a fraudulent campaign.
I'm waiting to hear from google...