pull down to refresh
0 sats \ 0 replies \ @k00b 5 Mar 2024
I like hearing about this kind of thing. I imagine large teams of engineers specialized like some once were in building castle walls. This sounds much like WAF with:
- protecting privacy when models are trained on sensitive data (no one should be doing that, right?)
- detecting when prompts are trying to get the AI to say something offensive
reply