[KMnO4]

I know that’s tongue in cheek, but I’d really appreciate having a NSFW text filter.

Some of my CX team is getting harassed by trolls who send profane messages and racial slurs through the Zendesk Chat interface. Zendesk does not have a way to filter out profanity (for Chat at least; there are options for emails or tickets).

I’ve found simple client side extensions that can censor words, but a better approach would use NLP to grasp the context.

“I hope you die” contains the same words as “My <product> got wet, I hope it didn’t die”, but with vastly different intention.

[dexen]

Google Jigsaw might have what you are looking for, under the "Harassment" headline [1]. Note however both Google (Alphabet, really) and the tool specifically, were accused of various forms of bias [2] by various sides of discourse.

--

[1] https://jigsaw.google.com/issues/

[2] for example https://boingboing.net/2019/08/14/white-aligned-english.html