There's a video going around that made a little bit of a stink... about how Twitter would share Trump's private messages, a little too gleefully.  Personally, this is the weakest of the attacks against Twitter to me. Sure, they have someone a little too enthusiastic sounding about violating someone else's privacy -- so there's some reasons for concern. But the basics of them complying with the law, and sharing everything with the DOJ, is not just a good idea, it's the law. Where I would have fished more, is whether they would have the same enthusiasm about revealing Obama, Hillary or Elizabeth Warren's private information, and that there was a sense of fairness in their creepy oversharing -- or whether it was partisan. And I'd rather they defended customer privacy more (by encoding things in a way where they can't mine the users private messages). But still, this is irksome at worst.
And the idea of automating this banning with AI and ML, doesn't make me feel any better. An algorithm is only as good and unbiased as the coder... and ML is only as good as the data that's pumped into it. It's easy to write an ML algorithm to think that "Big government is good", and "more liberty is bad". Or other far left intersectional hypocrisy. As I have no confidence in bay area folks ability to be non-partisan and objective (I live and work among them), thus I have no confidence in their FakeNews filtering to be anything by pro-fascism or pro-collectivist claptrap.
Another variant of this, is that they have a blue checkmark for those verified to be who they claim they are. But they'd pull it from people they don't like (like Richard Spencer) -- as if they're less sure of their identities? I'm no fan of Spencer or his ideology, but it's retarded to claim he's not validated as being who he claims to be. And Twitter of course only seems to pull this shit with some fanatics, and not others. (Show me a far left fanatic, where they did the same stuff). 
Banning anti-Hillary tweets
They show that they don't have the same standards of use, depending on if you're conservative or liberal/progressive.
Like they refused to smack Rosie O'donnel's bad behavior with the same enthusiasm they do conservatives. At least until they were shamed into behaving consistently (days later). 
Just like they destroyed Julian Assange's Wikileaks account, but then restored it, and refused to answer comments as to why. (And breaking followers links and posts in the process). 
And they purged Richard Spencer account, not really for violation of terms, but because they really didn't like who he is, or what he stands for.
They purged Milo Yiannopoulos account not for anything he did, but because of what they claim some of his followers did without any direct calls for attack by Milo. While the victim (Leslia Jones) behaved worse, and got no punishment. 
- Laura Loomer - De-verified
- British anti-Islam activist Tommy Robinson - De-verified
- Baked Alaska - blocked
- Shadow banning: https://www.projectveritas.com/2018/01/11/undercover-video-twitter-engineers-to-ban-a-way-of-talking-through-shadow-banning-algorithms-to-censor-opposing-political-opinions/
- No Check for you: https://hotair.com/archives/2017/11/16/twitters-blue-checkmark-purge-leads-questions-twitter/
- Anti-Hillary: https://www.infowars.com/twitter-admits-to-blocking-anti-hillary-tweets-during-2016-campaign/
- Assange: https://www.cbsnews.com/news/julian-assanges-official-twitter-account-not-appearing-wikileaks/
- Richard Spencer: https://www.theatlantic.com/politics/archive/2016/11/twitter-censorship-will-only-empower-the-alt-right/507929/
- Real Threat on Net Neutrality: http://www.breitbart.com/tech/2017/11/24/bokhari-google-facebook-twitter-and-youtube-are-the-real-threat-to-net-neutrality/