Hey! I'm fiddling around with this right now for a submission. What I've found so far makes it look like hate speech is not easy to catch:

1) Slang evolves.

2) Combining two languages, especially under-resourced languages, mean that a chunk of words fall outside the range of a hate-speech lexicon.

So maybe in languages which have huge resources devoted to it (English), its easy to figure out, but I don't know how well covered the evolving edge, and the code-mixed (language mixed) edges are.

Facebook is working on this.[1] There's even a $100,000 "Hateful Memes" competition.[2]

In order for AI to become a more effective tool for detecting hate speech, it must be able to understand content the way people do: holistically. When viewing a meme, for example, we don’t think about the words and photo independently of each other; we understand the combined meaning. This is extremely challenging for machines, however, because it means they can’t analyze the text and the image separately. They must combine these different modalities and understand how the meaning changes when they are presented together.

Facebook is going for the really hard case, where non-hate images and non-hate text combine to induce hate. They already have a text-only system.

[1] https://venturebeat.com/2020/05/12/facebook-is-using-more-ai...

[2] https://ai.facebook.com/tools/hatefulmemes/

If the CCP can do it —and they do it to catch all the circumventions Chinese internet users use to try to avoid censorship —which they do successfully, then we know it’s possible and if it’s possible the big players are doing it too.

The CCP doesn't manage to catch all circumventions, AND they are not a multinational firm crossing multiple jurisdictions.

So when we say big players, do you mean FB or do you mean the National Govs?

Side note - the CCP also employs manpower to handle the task, so its not just ML, my point hovers on the ML aspect of it.

The CCP doesn't need to catch them all. A 90% filter with a terrible false positive rate still accomplishes their societal engineering objectives.

People can be as hateful as they want and the government isn't coming after them. What people that use hateful verbiage are really arguing for is free from consequence speech and that just doesn't exist. They want to be allowed to be hateful and hurt Google's reputation and Google just needs to shut up aka these folks don't really value freedom of speech when people start using theirs to counter them.