Comment by vinnythejinny

Have you tested it ? I suggest you test it with multiple screenshots (literal scam text, just a normal conversation (not a scam) and borderline doubtful conversation. The AI is pretty good at identifying them, with a confidence score and risk level.