Stable Diffusion Safety Checker Question

Hi, I’m planning to use a stable diffusion model in an application with end users. Is the safety checker for Stable Diffusion 1.5 sufficient for preventing unfavorable outputs, or is Stable Diffusion XL and it’s safety checker more appropriate for use in production? Also, are there any other techniques that people have been using to further prevent unfavorable outputs, such as text input filtering? Any tips would be helpful, thanks!