Scientists developed an AI monitoring agent to detect and stop harmful outputs
Submitted by Anonymous (not verified) on Mon, 11/20/2023 - 19:30
The monitoring system is designed to detect and thwart both prompt injection attacks and edge-case threats.