Mmmm, are we sure that this is what Firefox 3 is doing right now? The sites you're linking to provide information about how the mechanism works in 2.x, not 3.x, and I don't see they were updated for the new version, maybe because the feature works in the same way in 3.x, or maybe because they're lazy to update it.
In any case, I think it's an acceptable compromise, the URL is only submitted when you encounter a phishing site, and frankly, in the case of these kind of pages, I don't see a privacy concern. It would be a good idea to submit only the domain, but the URL is also OK when it comes to this (mostly because phishing sites usually don't have anything else than a front page).
But there's a problem. The safe browsing feature was also expanded to include malware-infested sites, and perhaps not many people would be happy that Google knows about their ventures on warez and/or porn sites. While the current way safe browsing works is much better than the Google-backed option in Firefox 2.x (which meant your entire history was seen by Google), the explanation given by Google in the wiki (that again, could not be correct with 3.x) is at least nonexistent.
"We need to collect these information to improve our statistics". What's the ultimate outcome of this? In which way is Google using the information users are submitting to improve the service they're providing?
"You don't have to contribute any info to the service if you do not want", but from what I understand, this is unavoidable if you encounter a "bad" site, as it's performed automatically whatever option you choose, well, I guess if you close the tab nothing is submitted then.
While it's the typical Google behavior, perhaps the Mozilla team should try to provide the technical details over how everything works, at least for the sake of information, which would be beneficial to clear a bit my perception that maybe Google ways are getting too much presence in the Firefox project.
