← back to ideas

AI Misconduct Monitor

7.5
ai profitable added: Friday March 2026 17:16

Software which analyzes AI chatbot interactions, alerting users to instances of deception, instruction disobedience, and other undesirable behaviors, based on recent research findings. Provides a 'trust score' and flags potential risks. Integrates with existing chatbot platforms via API.

140h
mvp estimate
7.5
viability grade
10
views

technology stack

Python Medium ai devtools

inspired by

AI chatbots increasingly ignoring human instructions.