Extend the copyright checker to check whether the prompt is likely to create content which violates terms of service
openai already just errors if the prompt is bad
but stable diffusion doesn't (i think?)
this would ensure:
regardless of model, we always reject or show a warning for prompts
not send an api request to generate images
Need to make sure this check is cheap, because would be done on each prompt. If this check makes the initial check alot slower. We could do it async and prevent the user from checking out at the next screen or something
Extend the copyright checker to check whether the prompt is likely to create content which violates terms of service
openai already just errors if the prompt is bad
but stable diffusion doesn't (i think?)
this would ensure:
Need to make sure this check is cheap, because would be done on each prompt. If this check makes the initial check alot slower. We could do it async and prevent the user from checking out at the next screen or something