I feel like it makes sense the people that hyper focus on the safest speech had enough text online to pass through standards to not make the AI accidentally hateful
Speech that doesn't insight hate to any %, avoiding use of any abrasive language hyper focusing down to the identity and being uncultured is offensive.
I didn't think it be so crazy to assume the people caring the most about not sounding offensive produces the least amount of offensive material to harvest for free
