Toxicity detection algorithms in proactive moderation systems raise concerns about contextual complexities, inequalities, and potential misuse.
