How does one build this stuff. What kind of dataset?.. I bet it’s reddit.
These guys built an AI for username moderation (blocking peoples choice in name).
Though it does sound useful, I recall a time in which “linux” was a banned word on Microsoft’s connected MSN and/or XboxLive platform.
An API that can do the blocking automagically, where their biases are unknown, sounds like a good way to empower secret police decisions based on names.