The easiest way to combat this is to put your model behind an API and filter queries (midjourney, OpenAI) or just not make it available (Google). The tradeoff is that you're paying for everyone's compute.
I guess SD is betting on saving $ on compute being more important in this space than the ability to gatekeep certain queries. And the tradeoff is that you need to do nsfw filtering in your released model.
It will be interesting to see who's right in 2 years.
Making a bulletproof filter is incredibly difficult, even more so in a topic where image descriptions are written in a culture that often has to circumvent text filters. Both midjourney's and OpenAI's filter works mostly because of the threat of bans if you try to circumvent them. I'm not sure I would describe that as "the easy solution"
I guess SD is betting on saving $ on compute being more important in this space than the ability to gatekeep certain queries. And the tradeoff is that you need to do nsfw filtering in your released model.
It will be interesting to see who's right in 2 years.