He obviously doesn’t have the technical terminology down, but what he’s trying to say does make a little sense. If I understand correctly, he’s trying to say that they do collect the IP address (e.g. 208.80.152.201, which is wikipedia), but that they don’t collect urls (e.g. https://en.wikipedia.org/wiki/Invisible_pink_unicorn). Therefore they don’t record exactly what you’re reading, just where you’re getting it from.
(Of course, thinking persons realize that capturing the IP addresses is egregiously invasive, but I can see how it’s slightly less egregiously invasive than collecting URLs).
sigh but that is even more inane–given the amount of shared hosting, CDNs, and virtual hosts (not to mention all the other shenanigans actual criminals get up to) then just collecting the IP is worthless.