I have assumed for some time that de facto Gmail leaks information about the contents of my email. Having just read a related question, I'm wondering if my assumption is correct.
Suppose, for example, Merck buys an ad that links to keywords related to embarrassing medical condition XYZ. They link the ad to a URL in which they have encoded XYZ. When I click on the ad, whose text is something innocuous, I'm sent to the XYZ URL, whose contents sets a cookie saying "Gmail tagged as XYZ". Voila, Merck has now pried personal info from my Gmail account. (Merck also takes the precaution of logging my IP.)
Is there any reason this wouldn't work? Does anyone do this? The only defense I can see if Google wants to prevent this is to carefully examine keywords (nearly impossible to do with 100% efficacy) and also do some robot testing for cookies. Or legal defenses, but suing your clients gets messy.
(Note 1: Merck is a trademark and simply an example; I have no beef with them. Nor do I have any embarrassing medical conditions for which they sell pharmaceuticals. :)
Note 2: This is far more insidious these days, since Gmail stopped only displaying keyword ads by the particular email containing the keywords. Now you could click on an ad for a free iTunes song that leaks info about an email you read 10 minutes ago.)