AI-written message check
/ai check estimates whether a message was written by an AI. What it looks at and how much to trust it.
Usage
/ai check message_id:<id> (admin) reads the message and replies with a score, a confidence and a one-line reason. Get the ID with Developer Mode → right-click → Copy Message ID.
How it decides
It is a word and structure heuristic, not a call to an AI detector: it looks for vocabulary and phrasing that language models overuse, sentence uniformity and similar signals. Very short messages return "Text too short to analyze accurately" with a zero score. If it finds little of anything, the text is treated as probably human.
How to use the result
Treat it as a hint when deciding whether an application, appeal or review was copy-pasted from a chatbot, and ask a follow-up question rather than acting on the score alone. It cannot identify which AI, and it produces false positives on formal writing.
Unrelated to the AI Chat module, which makes the bot reply in channels and tickets.