Can AI check itself for safety and secrecy? Can I split prompts?

Thanks for your response @m-a.schenk I´ve been learning a lot with your posts.

I agree, a lot of sensitive data exchanged. I belive that those security checks are a challenge, but not impossible (right?).

In my mind, I imagine two separate boxes. One box have the GTP-3 “mind” and other deals with sensitive data in a close loop, both the inputs and the outputs could be encryped perhaps?

2 Likes