@jimmysmitty:
"thing that happen on GMails back end does not get seen by you, your system or your AV."
First, Gmail can be used either with a browser-based client or with an email client such as Outlook. I currently do both. In either case, what gets delivered to the client's computer is an HTML page. One (the browser-as-client version) is contained inside a larger HTML page. The other (the fat-client version) is embedded within the message contents as text/html within a multi-part MIME block. Go to w3c.org to learn more about these. In either case, compiled code running inside an internet browser processes the HTML or HTML segment, including the IMG tags that point back to the images that the marketers are transmitting. With fat clients, you can usually configure the software not to retrieve the image unless it comes from a trusted location. With thin (browser) clients, this is controlled by the application software running on the web server at Google. In order for this feature to do what Google says it does, they need to parse through your email message, locate any and all IMG (and similar) tags, retrieve the indicated images, store those on their servers, and *** modify the IMG tags in your email messages to now point at the images stored on their servers *** . Then, your client - be it a browser or a fat client - retrieves the image according to the URL in the message content, which now happens to point to Google's servers.
The one thing this does that you probably DONT want is that it validates your email address. Do not underestimate the significance of that. If no one ever downloads a target-identifying image in a marketer's email message (e.g. "IMG src="http
/blah.com/image.gif?email=you@email-address.com"), then they don't know whether that message ever went anywhere. It could be invalid. But if the message DOES get downloaded, then they know it went somewhere. Hopefully, Google - realizing this - is downloading the images for ALL messages that they receive, and is simply not storing the ones that go to invalid addresses. That would truly flummox the marketers because they would have absolutely no way to determine whether an address was valid or not, and every attempt to do so would simply put more load on their servers.