Comment by pomatic

The open source model seems to be uncensored, lending weight to the separate filter concept. Plus, any filter needs to be revised as new workarounds emerge - if it is baked in to the model that requires retraining, whereas it's reasonably light work for a frontend filter.