I think the fundamental difference is that spam detectors are seen as acting on their users behalf. If I say an email isn't spam, then it isn't and that's the final word on the matter.
Fake news classifiers don't give one classification per user. They don't try to predict for every single user whether that user would find it credible. Instead, they output a single value for the whole community. They act on the community's behalf. Because of this, some people will always disagree, and they will feel that the system is imposing on them. That's why it's political when spam detectors are not.
While you are right that the "ground truth" of spam vs ham comes from you, the process of classifying your mail relies on a lot of data provided by the community.
The story, like all other stories, is completely absent from google news. The stories announcing Michael Moore's documentary were timed (fine, not for sure purposely), to coincide with the 2nd day of O'Keefe videos. Upon visiting trends.google.com, James Okeefe never even made the list. You can scroll down through the top several hundred queries, and it wasn't in the list at all. The #1 story was Michael Moore anti-Trump doc for the past 24 hours. However, if you clicked on the michael moore link, and click compare, and start to type in james okeefe, and click on the autocomplete, it shows the true story. James okeefe stories dominated those about michael moore for the entirety of the past 24 hours. If you click on the michael moore story (in the 3 spot in my screenshots with sports above it), you can see that search interest hits 100 at 5pm, meaning, there is nothing bigger then it. However, once you add in james okeefe stories manually, you realize it was fudged. The day before that issue didn't exist, but starting after the first day of the james okeefe undercover videos, all the big pro trump/anti hillary stories never ever made the trending list through the election.
In summary James okeefe videos werent trending at all, got totally hidden from the youtube trending page, every single day, along with just about every big pro trump video, and literally hundreds of hashtags were hidden from twitter. I will link some screenshots below. If you think they are somehow manipulated, they are taken from a 20 minute video of me explaining how to see all of this for a noob. If you were actually interested, please just say so, so i can re-upload to an anonymous youtube account (its private right now, was meant as an instruction for a friend in the media to replicate the comparison for news.) I didn't compare with the wrong michael moore, they are all from the same exact time, etc. Like i said if you really think i am messing with you, i will upload the video this afternoon.. However I am assuming this will just get downvoted and ignored, but who knows. P.S. I know it doesn't say 'Michael Moore, Donald trump' as it does in the trending list, but the absolute figures correspond exactly, the ui just doesn't display the full combination of queries on the details page's chart title, again in the video if you don't believe. http://imgur.com/a/TX0T2
Even if it's not easy to know algorithmically, it is very easy to know a spam message when one sees the message.
News on the other hand, especially on political issues with differing opinions, are not that clear. And the underlying facts are not that major a factor either, because journalists can lie with statistics, be selective, omit things, put a false spin on the impact of real facts, etc.
And even who measures the "facts" can be political (e.g. a researcher paid by some mega-corporation, a government agency that wants to make the government look good, etc), and seemingly 100% accurate stats (e.g. on the "level of unemployment") cannot always be trusted and sometimes differ to reality by a huge margin (e.g. the level of unemployment can be manipulated in 100 different ways, based on the definition on unemployment, and various countries play all kinds of tricks with that.