Recently I had to permit the term of "ow" when someone wanted to simply type "How are you <3" on a stream. This rather baffled me. I have since found similarly harmless words have triggered very low levels of automod sensitivity despite being rather simple words. Someone on reddit thankfully screenshotted some examples of the settings being used for the How are you <3 instance and has uploaded them for easy viewing:

Settings: https://i.imgur.com/3vLFx3Z.png
Results: https://i.imgur.com/VYjp4Sj.png

I can also state that I've had to permit the word 'dinosaur' on the same levels of sensitivity. I think it AutoMod should be more carefully reviewed. Clearly globally offensive words such as racist (or other) slurs should be included in discrimination, but not words such as 'dinosaur' - at least, not at the lower settings. Perhaps at the highest setting you could put something like this in. For the "ow" note, it appears people on Twitter have had similar experiences with the letters "on" and "op" within other words. I've not verified their claims but I have no reason to expect they aren't factual.

Could we please have AutoMod sensitivity levels adjusted in the following ways:

1) Adjust appropriate levels for words such as 'dinosaur' which feature in simple inoffensive discussion more regularly than they do as slurs.

2) Remove the conditions set for letters within other words such as ow, op and on (seemingly when including a heart emote after a space in the same sentence). I understand that people regularly try to get around word filters by being creative, and that AutoMod probably tries to combat this with particular algorithms, but when it is considerably more likely to impede on innocent discussion than offensive statements then it is probably something that should be left to genuine streamers and moderators to manage. Especially if it interrupts frequent statements such as "How are you <3"