Re:Prompt injection attacks against GPT-3
by sinanuozdemir on Tuesday September 20, 2022 @10:14AM
(#32911918)
One idea I have is to use a secondary system like grammarly to check if the input is "imperative" in nature meaning it's formatted to command the AI to do something. It's pretty naive but could be a simple start.
[ Parent
| on HN ]
There are no comments on this comment yet.