Run threat assessments through multiple AI models and log where they disagree....

https://500px.com/p/graveyardwindslvlrf

Run threat assessments through multiple AI models and log where they disagree. Write down every assumption, name the person who makes the final call, and set a follow-up review date for each case